Examining the Impact of Entrepreneurial Orientation, Self-Efficacy, and Perceived Business Performance on Managers’ Attitudes Towards AI and Its Adoption in Hospitality SMEs

Abstract

In the competitive hospitality sector, the adoption of Artificial Intelligence (AI) is essential for enhancing operational efficiency and improving customer experiences. This study explores how key entrepreneurial traits—Entrepreneurial Orientation (EO), Entrepreneurial Self-Efficacy (ESE), and Perceived Business Performance (PBP)—influence managers’ attitudes toward adopting AI in small- and medium-sized enterprises (SMEs). Ts research utilizes data from 287 respondents, gathered through field research with a survey designed to measure the relationships among constructs, employing structural equation modeling (SEM) for analysis. Results reveal that PBP and certain ESE dimensions, such as Initiating Investor Relationships and Developing New Products, have only a modest positive effect on AI adoption. In contrast, EO—specifically Proactiveness and Innovativeness—exhibits a weak negative impact. Importantly, none of these factors directly affect managers’ attitudes toward AI. Instead, this study highlights that managers’ positive attitudes are the strongest predictors of AI adoption, aligning with the Technology Acceptance Model (TAM). The findings offer new insights into key entrepreneurial factors driving AI adoption and emphasize the need for targeted education and supportive policies to facilitate AI integration in hospitality SMEs. Fostering a positive perspective on AI is more important for adoption than overcoming skepticism, as negative attitudes do not influence AI adoption.

1. Introduction

Tourism is a major driving force for the European Union’s (EU) economy, accounting for 10% of its Gross Domestic Product (GDP) [1]. In the Republic of Slovenia, a smaller EU economy, tourism also plays a crucial role, contributing 9.2% to the national GDP. Notably, 99.8% of businesses in Slovenia are small- and medium-sized enterprises (SMEs) [2]. Recognizing the vital role of SMEs, the EU Commission has prioritized advancing Artificial Intelligence (AI) within these businesses to create a robust AI ecosystem across the EU, fostering economic growth and innovation [3].

AI refers to the creation of intelligent systems capable of understanding and interacting with their environment [4]. This technology opens new avenues for innovation, communication, and collaboration, but it also brings complex ethical, legal, and social challenges [5,6]. As AI evolves, its impact on communities and businesses will require proactive engagement with this rapidly advancing technology [7,8].

In the hospitality industry, traditionally slow to adopt new technologies due to its labor-intensive, customer-centric nature [8], AI offers several compelling advantages for small hospitality businesses, such as enhancing operational efficiency [9], improving inventory, financial, and strategic management [10], elevating customer experiences, and supporting sustainable practices [11].

Despite these benefits, SMEs face unique barriers to AI adoption compared to larger enterprises. Common obstacles include outdated infrastructure, limited IT skills, and high costs [12,13]. Organizational factors like flat structures, limited IT expertise, lack of management support, data security concerns, ethical issues, and cultural barriers further complicate the adoption process [11,14,15]. According to Schwaeke et al. [16], the existing literature reveals a fragmented landscape that hinders our understanding of how SMEs use AI [16].

To better understand AI adoption in SMEs, recent studies have examined how managers’ personality traits influence this process, reflecting an evolving understanding of adoption dynamics [8,17,18]. Research indicates [7,17,18,19] that various personal characteristics of managers—such as self-determination, trust, fear, cultural background, innovativeness, adaptability, openness to change, generativity, AI anxiety, and AI agreeableness—can significantly influence their perceptions, attitudes, and use of AI. Specifically, attitudes play a crucial role as a precursor to information technology and AI adoption, influencing users’ intentions and behaviors. The Theory of Planned Behavior (TPB) and Technology Acceptance Model (TAM) both emphasize attitudes as key determinants in this context (further discussed in Section 2.3) [20].

In this study, we select Entrepreneurial Orientation (EO), Entrepreneurial Self-Efficacy (ESE), and Perceived Business Performance (PBP) as independent variables to explore their influence on managers’ attitudes toward AI adoption. EO is crucial as it reflects a firm’s capacity for innovation and adaptability, essential in a rapidly evolving technological landscape [21]. ESE captures managers’ confidence in their abilities [21], potentially influencing their openness to adopting new technologies such as AI. PBP offers insights into managers’ subjective assessments of their businesses’ performance [22], which may affect their willingness to adopt innovations based on perceived operational benefits.

However, to the best of the author’s knowledge, there is a gap in the literature regarding the impact of key entrepreneurial traits, such as EO and ESE [21], on managers’ attitudes toward AI and its integration into existing business processes. Recent research [22,23,24] also emphasizes PBP’s role in influencing SMEs’ overall performance. These studies suggest that EO, ESE, and PBP play important roles in decision-making, with significant impacts on SMEs’ outcomes. Therefore, it is essential to investigate these three constructs further, as they may (in)directly affect AI adoption through their influence on managerial attitudes.

Accordingly, we formulate our main research question (RQ) as follows: How do key entrepreneurial traits such as EO and ESE, along with PBP, influence managers’ attitudes toward AI and its adoption in hospitality SMEs?

Given the potential for new technologies to enhance SME performance, this study aims to address this gap by examining how key entrepreneurial traits, along with PBP (as independent variables), influence managers’ attitudes toward AI and their intention to incorporate it (both as dependent variables) into operational processes within hospitality SMEs. To explore the relationships among these variables, data from 287 Slovenian hospitality SMEs will be analyzed through factor analyses and structural equation modeling (SEM) based on the research model presented in Figure 1.

This research contributes to the existing body of knowledge by examining the intersection of entrepreneurial traits and AI adoption in hospitality SMEs, emphasizing the pivotal role of positive managerial attitudes as key drivers. Theoretically, this study enriches discussions on AI adoption by integrating key entrepreneurial constructs such as EO, ESE, and PBP, while highlighting their limited direct impact on AI adoption behavior. Practically, the findings provide valuable insights for policymakers and industry practitioners, advocating for targeted educational interventions to foster positive managerial attitudes towards AI. These positive attitudes play a critical role in overcoming adoption barriers and driving the digital transformation of SMEs within the hospitality sector.

This paper is structured as follows: Section 2 discusses the theoretical framework and hypotheses, Section 3 details the research methods, Section 4 presents the data analysis, and Section 5 and Section 6 cover the findings, implications, limitations, and suggestions for future research. Our study not only offers practical insights for tourism businesses and policymakers to enhance AI adoption but also contributes new knowledge to the academic field. This is especially relevant now, as AI in hospitality and tourism is still in its early stages of adoption [25].

2. Theoretical Background and Hypotheses

2.1. AI and SMEs’ Performance

The EU AI Act, adopted in 2024, defines AI as “a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments” [26].

Since AI is a novel research area, AI tourism research has grown exponentially over the last decade, with a focus primarily on the hotel and travel sectors [11,12,14,27,28]. Literature reviews have identified five major types of AI applications in the hospitality sector [28,29]: search/booking engines, virtual assistants/chatbots, robots and autonomous vehicles, kiosks/self-service screens, and augmented and virtual reality (AR/VR) devices. These applications are utilized across four key areas: forecasting, operational efficiency, guest experience, and sustainability [11].

Even though the adoption of AI is still in its early stages [25,30], studies indicate that SMEs implementing AI report significant benefits [31,32,33]. AI applications help SMEs facilitate data analysis, receive data-driven decision support, and enhance overall operational efficiency [34], which can improve financial performance [9]. Additionally, AI implementation can reduce human errors and elevate customer satisfaction by minimizing waiting times [35].

Conversely, concerns have been raised about the adverse effects of AI adoption [36]. Issues such as AI opacity and perceived employee risks have been shown to impact organizational performance negatively [37]. These challenges can potentially lead to increased employee turnover and decreased job satisfaction [38].

Nevertheless, AI is regarded as a driving force in the “fourth industrial revolution”, presenting opportunities to enhance efficiency and accuracy. However, its successful implementation requires a robust strategy to tackle organizational and ethical challenges [26]. In the EU context, the report Prospects for the EU 2024 [39] highlights the EU’s competitive disadvantage in AI compared to major players such as China and the USA. The report underscores the necessity for increased investment and advancements in cutting-edge technology to close this gap.

2.2. AI Implementation Challenges

Despite the transformative potential of AI, its adoption in tourism settings remains modest [35]. A study on digitalization in the EU restaurant industry reveals low implementation rates of AI in business processes [40]. Similarly, research on German SMEs indicates that managers are hesitant to shift from traditional technologies to AI, resulting in lower adoption rates [33].

The road to integrating AI is fraught with challenges, both environmental (external) and organizational (internal). Environmental barriers include financial constraints and technical limitations, which can be particularly difficult for SMEs to overcome [13,25]. Organizational hurdles involve difficulties such as integrating AI with existing processes within simpler organizational structures, addressing employee skills and training needs, and ensuring data security [8,41,42].

Recent research underscores the significance of internal organizational and cultural factors in AI adoption. Key elements influencing SMEs’ adoption of AI include employees’ work abilities, the socio-emotional well-being of family-run SMEs, entrepreneurial culture [34], and the level of management support [35,43,44]. Hospitality SMEs face additional challenges related to managerial perceptions of AI’s benefits and their customer-centric focus [30,45,46]. Many tourism SMEs are managed by their owners [22], placing the owner–manager at the heart of decision-making. These studies highlight the intricate psychological and social dynamics that influence how SMEs approach and implement AI technologies.

In summary, while AI offers numerous benefits, such as improved data analysis, automation of repetitive tasks, and enhanced customer experiences, its successful implementation in SMEs requires overcoming several significant barriers [36]. Addressing these challenges is crucial to unlocking the full potential of AI in the tourism and hospitality sector [47].

2.3. Models for AI Adoption Analyses

When it comes to understanding how and why users accept AI, researchers have developed several models that examine various environmental and organizational factors. Some of the most well-known models include the Theory of Planned Behavior (TPB), Davis’s Technology Acceptance Model (TAM), the Technology–Organization–Environment (TOE) framework, and the Unified Theory of Acceptance and Use of Technology (UTAUT) [48]. In recent years, new models have emerged to address the specific nuances of AI adoption. For instance, the AI Device Use Acceptance Model (AIDUA) and the Technological Transformation Model [49] bring new insights into how users interact with AI technologies.

Several key models presented above, such as TAM, TPB, and UTAUT, include attitudes (discussed in more detail in the section below) as critical indicators of actual technology adoption. These models aim to explain and predict how users accept and utilize new technologies, with attitudes being central to the process. Among these models, TAM has become particularly popular in studies involving SMEs [50,51,52,53], even though it was originally developed to explain technology adoption from the user perspective. This model emphasizes perceived ease of use (PEOU) and perceived usefulness (PU) as key determinants of user attitudes, which in turn influence technology adoption [51].

In particular, the core constructs of TAM align closely with the goals of our study. They allow us to explore how managers’ attitudes toward AI drive implementation. The focus of TAM on personal attitudes and its previous use in SME research makes it a suitable framework for our investigation. While attitudes are essential for predicting IT and AI adoption, some researchers (presented below) have expanded TAM by incorporating additional critical variables that address unique challenges influencing AI adoption behavior. These extended models incorporate factors such as social influence, performance expectancy, and facilitating conditions in the adoption of AI technologies [51,54,55]. Key modifications also include integrating user trust in AI and perceived quality of AI outputs, which are critical for user acceptance in AI-driven applications [56]. Additionally, extended models consider functional and motivational aspects, where hedonic and social motivations significantly impact users’ continued trust and intended use of conversational AI [57]. Empirical research has validated these modified models across a variety of applications, demonstrating their effectiveness in predicting user acceptance [58,59]. Furthermore, longitudinal models emphasize feedback mechanisms, showing that actual use can enhance continued intended use—an often-overlooked factor in the original TAM [57].

In our study, we specifically use TAM as a framework for understanding the relationship between attitudes and AI adoption. Since recent research indicates that several variables influence attitudes (in our study, EO, ESE, and PBP are investigated), these additional constructs provide a more comprehensive model for analyzing AI adoption.

This approach aligns with our focus on attitudes as central to AI implementation and supports our aim of examining how attitudes, influenced by these additional variables, drive AI adoption. This perspective reflects the growing consensus among researchers [47,59] that adaptable, sector-sensitive models may be necessary to effectively analyze AI acceptance in dynamic industries.

2.4. Attitudes Towards AI

An attitude reflects a person’s long-term evaluation, emotion, or stance toward something [7]. Attitudes significantly influence how individuals intend to use and engage with technology. Research has consistently highlighted the importance of attitudes in AI adoption, given the diverse forms AI can take and its varied interactions with individuals [60]. A systematic literature review by Kelly et al. [7] revealed that attitudes are a powerful predictor of the intention to use AI, making them a key variable in explaining the ‘attitude–behavior’ link in the AI adoption process.

However, attitudes towards AI can be influenced by various psychological factors. For example, Schepman and Rodway [20] found that introverted individuals tend to have more positive attitudes toward AI usage. Similarly, Park and Woo [61] found that personal innovativeness in information technology strongly correlates with more positive attitudes towards AI. These findings highlight how individual psychological traits can shape perceptions and acceptance of AI technologies.

On the negative side, research indicates [61] that individuals with higher levels of neuroticism often have more negative views of AI; this was supported by demonstrating that personality traits such as agreeableness, AI configuration anxiety, and AI learning anxiety are significant predictors of negative attitudes toward AI. Sindermann et al. [62] further revealed that neuroticism is linked to a greater fear of AI. Additionally, Bartneck et al. [63] observed that support for stringent AI regulations is related to agreeableness and neuroticism but inversely related to openness to experiences. This indicates a complex interplay of personality traits in shaping individual attitudes toward AI.

Furthermore, demographics play a significant role in shaping attitudes toward AI. Research indicates that men, individuals with higher levels of education, and younger people typically exhibit more positive attitudes toward AI [20]. In contrast, personal barriers such as skills gaps, religious beliefs, and resistance to change—often embodied by the mindset of “we have always done it this way”—can negatively impact AI attitudes [32].

In theory, the connection between attitudes and technology adoption is well-established. However, to the best of the authors’ knowledge, no studies have investigated how attitudes, in the AI adoption process (see Figure 1), are potentially influenced by SME managers’ EO, ESE, and PBP.

2.5. Entrepreneurial Orientation (EO) and Entrepreneurial Self-Efficacy (ESE)

EO is defined as entrepreneurial inclination towards innovation, proactivity, and risk-taking. It encompasses the policies and practices that guide entrepreneurial decision-making and actions within a company [64]. EO is typically analyzed according to its three core dimensions: Innovativeness, Proactiveness, and Risk-taking (see Appendix A). Essentially, EO captures a company’s willingness to embrace new ideas, anticipate future trends, and take calculated risks to maintain a competitive edge in the market [18]. Moreover, EO has also been shown to be significant for SMEs’ green practices, leading to improved environmental performance and a competitive advantage in the marketplace [65]. In contrast, ESE refers to an individual’s confidence in their ability to handle various entrepreneurial tasks. This belief affects how individuals think, feel, and act, significantly impacting their entrepreneurial intentions and behaviors. ESE encompasses several dimensions (see Appendix A), including creating new products or services, fostering an innovative environment, attracting investors, defining core purpose, coping with unexpected challenges, and managing human resources effectively [24].

Both EO and ESE have been extensively studied and are known to positively influence the overall business performance and competitiveness of SMEs [22,64,66]. Despite their significance, there appears to be a gap in research regarding how EO and ESE impact AI adoption within hospitality SMEs. Given that EO and ESE are crucial psycho-sociological factors influencing entrepreneurial behavior [21], it is plausible that they might also play a key role in the successful adoption of AI.

Specifically, managers who exhibit high levels of innovation, proactivity, and risk tolerance, along with those who have strong confidence in their abilities, are likely to adopt and leverage new technologies more effectively [64]. This connection is vital for enhancing the growth and competitive edge of hospitality SMEs in the technology-driven landscape. Therefore, we propose the following hypotheses (see Figure 1):

H1. 

EO positively influences managers’ attitudes toward AI and, consequently, its adoption.

H2. 

ESE positively influences managers’ attitudes toward AI and, consequently, its adoption.

2.6. Perceived Business Performance (PBP)

PBP focuses on how managers assess their own SMEs’ financial performance and the quality of their offerings. This self-assessment approach is quite common in research [22,23,24]. Studies suggest that how managers perceive their business’s performance can significantly impact SMEs’ overall performance through effective managers’ decision-making [22,64]. Specifically, high-quality services and the achievement of organizational goals are often linked to management practices informed by managers’ realistic perceptions of their business’s performance [22].

Additionally, Sharma et al. [67] have shown that management support is crucial in driving AI adoption, which can enhance SMEs’ business performance. Therefore, managers need to accurately perceive their SME’s performance, as this influences their business decisions and strategy. This leads us to our third hypothesis (see Figure 1):

H3. 

PBP positively influences managers’ attitudes toward AI and, consequently, its adoption.

To provide a comprehensive overview of the literature that guided our hypotheses development,

Table 1. Summary of key literature and aspects considered in hypothesis development.

Citations	Key Aspects Considered
[21]	Influence of EO on innovation and risk-taking in SMEs.
[24]	Role of EO in fostering a culture of innovation in hospitality SMEs. Relationship between ESE and proactive management behaviors in SMEs. Link between PBP assessments and strategic planning in SMEs.
[64]	Correlation between EO and successful technology adoption. ESE’s enhancement of decision-making capabilities for adopting innovations.
[23]	Impact of ESE on confidence in adopting new technologies.
[22]	Effects of PBP on managerial decision-making and technology adoption.
[67]	Importance of PBP perceptions in forming management practices.
[11]	Positive managerial attitudes as predictors of AI adoption.
[60]	Influence of personality traits on attitudes toward AI.
[61]	Correlation between openness to change and positive attitudes toward AI.
[25]	Factors influencing low AI adoption rates.
[31]	Benefits of AI adoption for operational efficiency and customer satisfaction.
[32]	Barriers to AI adoption and the role of positive managerial attitudes in overcoming challenges.
[66]	For SMEs, EO is important because it enables survival and growth through innovation, proactivity, and effective management of limited resources in various market conditions.
[68]	Innovativeness positively impacts SME performance but not growth. Risk-taking positively impacts SME growth but not performance. EO is reflected in managerial behaviors and attitudes.

summarizes key studies and the aspects considered from each citation that underpin the constructs explored in this research.

3. Methodology

3.1. Instrument Design

The variables included in the research instrument were carefully selected and adapted from prior studies. Given the topic’s novelty and evolving nature, a detailed review of the recent literature from major academic databases was conducted. This review focused on tourism and SME research over the past five years, utilizing keywords such as EO, ESE, PBP, attitudes, AI, digitalization, hospitality, and SMEs.

EO and ESE were measured using established scales. Specifically, this study employed a 12-item scale for EO developed by Kraus et al. [69] and a 23-item scale for ESE from DeNobble et al. [70], as adopted by Planinc et al. [24]. These scales have been extensively validated in previous research and effectively capture the core dimensions of EO and ESE. PBP was assessed using a 9-item scale from Kukanja et al. [22], which has proven to be a reliable measure of managers’ self-assessed business performance. All variables were measured on a 5-point Likert-type ordinal scale, by which respondents indicated their level of agreement or frequency of occurrence.

Attitudes toward AI were investigated using the 20-item General Attitudes toward Artificial Intelligence Scale (GAAIS) from Schepman and Rodway [20]. This scale is designed to measure individuals’ attitudes toward AI, encompassing positive and negative emotions. It consists of two subscales that reflect positive and negative attitudes towards AI (see Appendix A). The response options for each item ranged from 1 (Strongly Disagree) to 5 (Strongly Agree), with reverse scoring applied to the negative emotion subscale.

AI adoption was measured using a scale that identifies major types of AI applications within the hospitality industry based on validated scale from previous research [28,29]. This scale included 11 items, and the intention to use AI was rated on a scale from 1 (I do not intend to use) to 5 (I intend to use/using). The category “I intend to use/using” combines those who actively utilize AI in their business processes with those who plan to start using AI in the near future. An additional sixth option, “Not familiar with this type of AI”, was included to capture respondents’ nonfamiliarity with specific AI technologies.

Finally, demographic information about respondents was collected, focusing on variables such as age, gender, formal education, years of experience in the industry, and ownership status [22]. Additionally, basic details about the SMEs were gathered, including the number of employees and years of operation, to provide context for this study’s findings.

This comprehensive approach to instrument design ensures that the measurements used are robust and relevant to the research objectives, capturing a nuanced understanding of the investigating constructs influencing AI adoption in hospitality SMEs.

Table 2. Summary of questionnaire constructs.

Construct	No. of Items
PBP	9 (P1–P9)
ESE	23 (S1–S23)
EO	12 (O1–O12)
Attitudes Toward AI	20 (Q1–Q20)
AI Adoption	11 (U1–U11)

Note: detailed research questions corresponding to these constructs can be found in Appendix A.

presents a summary of the constructs measured in the questionnaire used for this study. Each construct includes the number of items designed to assess specific aspects related to PBP, ESE, EO, AI attitudes, and AI adoption.

3.2. Data Analysis

Data analysis for this study utilized IBM SPSS Statistics 26 and IBM SPSS AMOS, following a structured approach to ensure the robustness and validity of the results.

Initially, descriptive statistics were calculated to provide an overview of the dataset, including measures of central tendency and variability. This step was crucial for understanding the general characteristics of the data and identifying any immediate patterns or anomalies.

Next, Exploratory Factor Analysis (EFA) was performed to uncover the underlying factor structure of the observed variables. EFA identifies the number of factors and the loadings of each variable on these factors, helping to determine the internal reliability of the variables. It simplifies the data structure by reducing the number of observed variables into fewer, more interpretable factors.

Following EFA, Confirmatory Factor Analysis (CFA) was used to assess whether the factor structure identified through EFA was appropriate and to validate the measurement model. CFA tests the fit of the hypothesized model to the observed data, ensuring that the model accurately represents the relationships between the variables (see

Table 3. Standardized factor loadings, validity, and reliability indicators (CFA).

Constructs	Variables	Factor Loadings (λ)	CR	AVE	Cronbach α
PBP	P4	0.72	0.93	0.59	0.90
	P5	0.71
	P6	0.86
	P7	0.80
	P8	0.78
	P9	0.74
Investor relationship (ESE)	S12	0.86	0.93	0.72	0.88
	S13	0.92
	S14	0.75
AI Adoption	U2	0.79	0.92	0.60	0.87
	U3	0.86
	U4	0.78
	U7	0.75
	U11	0.68
AI Attitudes	Q5	0.76	0.88	0.53	0.82
	Q7	0.73
	Q14	0.66
	Q18	0.73
Innovativeness (EO)	O3	0.73	0.91	0.68	0.86
	O5	0.90
	O6	0.84
Proactiveness (EO)	O10	0.67	0.87	0.58	0.80
	O11	0.78
	O12	0.82
Opportunity development (EO)	S1	0.76	0.90	0.64	0.84
	S2	0.82
	S3	0.82

Note: All factor loadings are significant at p < 0.05. Values for CR, AVE, and Cronbach’s alpha meet the recommended thresholds (CR > 0.70; AVE > 0.50; α > 0.70).

and

Table 4. Discriminant validity of factors.

Discriminant Validity			1	2	3	4	5	6	7
	Construct	Sqrt. (AVE)	0.769	0.847	0.774	0.724	0.826	0.759	0.798
1	PBP	0.769	-
2	Investor relationship (ESE)	0.847	0.423	-
3	AI Adoption	0.774	0.094	0.157	-
4	AI Attitudes	0.724	0.059	0.077	0.520	-
5	Inovativeness (EO)	0.826	−0.078	−0.159	−0.101	0.072	-
6	Proactiveness (EO)	0.759	−0.024	0.002	−0.055	0.158	0.656	-
7	Opportunity development (EO)	0.798	0.441	0.535	0.156	0.073	−0.092	−0.035	-

).

Finally, SEM was applied to test the theoretical model. SEM examines complex relationships among latent (unobserved) and manifest (observed) variables. It provides a comprehensive approach to assessing both the direct and indirect effects within the model, allowing for an in-depth analysis of the interactions and impacts among variables (see Figure 2).

3.3. Sample Description and Data Collection

The sample for this study consisted of SMEs based in the Republic of Slovenia, specifically categorized under the EU standard classification of activities (NACE) as I55 and I56, which encompass Accommodation and Food and Beverage Service Activities. According to data from the official business register [71], there were 8303 businesses in these classifications in 2023.

Given the complexity of tourism SMEs, which often engage in multiple business activities and are classified into various subcategories (e.g., I56.1—Restaurants and Mobile Food Service Activities), direct comparisons can be challenging. To address this, our study focused specifically on SMEs whose operational revenue was solely derived from I55 (Accommodation) and I56 (Food and Beverage Service Activities) and excluded those with franchise operations. We concentrated on SMEs with similar operational characteristics, such as those providing bed accommodation services (e.g., hotels, motels, and bed and breakfasts) and table service (e.g., inns and snack bars), excluding establishments like camps, mobile service units, and takeaway facilities. This approach ensured a more homogeneous sample and allowed for more accurate comparisons among facilities within the target sector. Due to the lack of detailed official information on the specific characteristics of SMEs in our study, we employed a convenience sampling method. It is important to note that this sampling method may limit the representativeness of the sample (see also Section 6).

Data collection was conducted from January to July 2023 by ten interviewers. The process started with pre-collecting data from public records to identify eligible facilities. Facilities that did not meet the inclusion criteria were excluded from further consideration. Respondents were required to confirm that their businesses derived no revenue from sources other than the I55–I56 activities. The reliance on self-reported information could affect the accuracy of the data, as respondents might not fully disclose all income sources. Consistent with similar research [8,22] the primary respondents were the managers and/or owners of these companies. In cases in which a facility did not meet the inclusion criteria or if the manager declined to participate, interviewers moved on to the next eligible facility. By the end of the data collection period, the sample included 287 SMEs, representing just over 3.46% of the total population in the I55–I56 classifications. While the sample size provides a solid basis for analysis, it may still limit the generalizability of the findings to the entire population of SMEs in these NACE categories.

4. Results

4.1. Sample Characteristics

The data reveal that 66% of respondents were men, and the majority had completed secondary education (57%). The age distribution showed that the largest group was between 46 and 55 years old (35%). In terms of experience in the hospitality industry, most respondents had been in the business for 21 to 30 years (31%).

Regarding SME characteristics, most businesses had been in operation for less than 10 years. A significant proportion are managed by individuals who are also the owners, indicating a strong entrepreneurial spirit (70%). Additionally, 61% of these SMEs are family-owned enterprises.

This study evaluated the values for various research items across the five constructs by calculating mean values (M) and standard deviations (SD). For PBP, item P1 (Our guests are satisfied with our products or services) received the highest mean score of 4.68 (SD = 0.50), indicating a strong positive perception of guest satisfaction. In the context of ESE, item S3 (I can identify areas for potential growth) was rated highest, with a mean of 4.25 (SD = 0.71), reflecting a high level of confidence among managers in identifying growth opportunities. For EO, item O8 (In our competitive environment, it is wiser to make conservative and incremental decisions) had the highest mean score of 3.23 (SD = 1.18), suggesting a tendency towards cautious decision-making in competitive scenarios. Among AI adoption variables, U1 (Smart booking platforms and/or online food ordering apps) was the most commonly adopted technology, with a mean score of 4.33 (SD = 2.01).

In contrast, the lowest scores were observed in different areas. For PBP, P6 (I am satisfied with the profitability) had a mean score of 3.79, indicating managers’ relatively lower satisfaction with profitability (SD = 0.86). The ESE item S7 (I can determine what the business will look like) scored 3.47 (SD = 0.97), suggesting a less confident perception regarding long-term business vision. For EO, item O12 (We usually wait for the leading competitor to enter the market first with new products and services) received a low mean score of 2.16 (SD = 1.18), reflecting a more conservative approach towards market entry. In terms of AI adoption, U10 (Augmented Reality—AR and Virtual Reality—VR applications) scored the lowest, with a mean score of 1.82 (SD = 1.22).

In terms of attitudes towards AI, the average score for positive GAAIS items was 2.70 (SD = 0.07), while negative GAAIS items averaged slightly higher at 2.97 (SD = 0.16).

4.2. Validation of the Model (EFA and CFA)

To validate the model, we first tested for normality of distribution. The Kaiser–Meyer–Olkin (KMO) measure was 0.822, and Bartlett’s test of sphericity was significant (p = 0.00 < 0.05; app. X² = 11,611.469; df = 2775), confirming the data’s suitability for EFA.

Accordingly, we proceeded with EFA using the maximum likelihood method and varimax rotation. Variables with low factor loadings (<0.6) were excluded, resulting in 27 retained variables out of 75, spanning seven constructs. This decision was primarily influenced by the sample size, as it plays a significant role in SEM, and it is crucial that the model is not overly complex. Therefore, we opted for a higher criterion, which is widely recommended in the literature [72]. Including more indicators in the model would have resulted in an overly complex structure given the sample size (see also Section 4.1 and Section 4.3).

It is important to highlight that, in designing the questionnaire, we followed an exploratory research approach rather than a confirmatory one. Consequently, we included more variables in the initial measurement model based on previous theoretical findings. Additionally, the goal of our study was not to validate an existing measurement instrument but to develop a new questionnaire for first-time measurement, contributing to the theory. When working with established psychological measurement instruments that have been thoroughly tested and validated, using as many original indicators as possible is crucial. In contrast, newly developed instruments focus on discovering meaningful constructs.

In principle, we could have included more indicators—setting the criterion at a 0.5 factor loading would have allowed for that. However, this would result in a more complex model, which would not be suitable given the sample size. Most researchers recommend using sample sizes of at least 200/5 or 10 cases per parameter [72]. In this context, the standardized factor loading for all items exceeded the threshold limit of 0.6 [73].

Notably, all variables related to the ESE dimensions Building on the Innovative Environment and Defining Core Purpose, as well as the EO dimension Risk Orientation, were eliminated from the analysis. The number of variables associated with the initial constructs of PBP, positive AI attitudes, and AI adoption was reduced. The initial dimensions of EO and ESE were redefined into new independent factor groups: Investor Relationship (ESE), Innovativeness (EO), Proactiveness (EO), and Opportunity development (EO). Each new factor group (constructs) consists of at least three indicators (see Table 3 and Appendix A).

Following EFA, CFA was conducted using the maximum likelihood method in AMOS. As shown in the table below, all indicators exceeded the 0.6 threshold, and the values for Composite Reliability (CR), Convergent Validity (AVE), and Cronbach’s alpha (α) were all above the recommended levels (Cronbach α > 0.80; AVE > 0.50; CR > 0.70) [71].

Additionally, discriminant validity was assessed to prevent multicollinearity issues. Discriminant validity determines whether the constructs in the model are highly correlated among them or not. A commonly accepted criterion for construct discriminant validity is that the square root of average variance shared between a construct and its indicators should be greater than the intercorrelations shared between the constructs in the model [74]. Table 4 presents the inter-construct correlation matrix, with the square root of the average variance extracted displayed on the diagonal. Good discriminant validity is demonstrated because all the diagonal elements are greater than corresponding off diagonal elements.

Convergent and discriminant validity are both considered subcategories or subtypes of construct validity, which is confirmed for all the constructs [72,73].

4.3. Structural Equation Model (SEM)

Next, we moved forward with SEM. The minimum discrepancy ratio (χ²) concerning degrees of freedom (dfs) was found to be appropriate (χ²/df = 512.84/318).

Table 5. Model fit.

χ2	df	p	RMESEA	CFI	TLI	PNFI
512.84	318	0.00	0.04	0.95	0.94	0.74

Note: χ²—Minimum of Discrepancy, df—Degrees of Freedom, RMSEA—Root Mean Square Error of Approximation (<0.05 or 0.08); CFI—Comparative Fit Index (>0.90); TLI—Tucker–Lewis Index (>0.90), PNFI—Parsimonious Normed Fit Index (>0.60).

shows multiple indexes of fit which were developed as an answer to a sensitivity of Chi-square statistics. As can be seen from table below, all indices confirmed a good model fit, indicating that the model is both well-fitting and parsimonious [73].

The SEM (Maximum Likelihood Estimation Method was applied) presented in Figure 2 follows a hypothesized model based partly on theoretical findings but also as a result of an exploratory approach (EFA), as theory is new to the field. Based on those results, CFA was performed to validate measurement instruments and to ensure that existing theoretical constructs are accurately represented in their data. As some indicators needed to be eliminated, EFA was crucial as a pretesting stage before CFA. The final SEM encompasses 27 observed variables and seven constructs, always considering sample size and trimming the constructs gradually and with a great deal of care to ensure validity and reliability of constructs used in the model, each supported by at least three indicators. This model collectively explains 31% of the variance in AI adoption.

Since not all variables and constructs (see Appendix A and Figure 1) met the criteria for inclusion in the analysis (as described above), we revised the model by creating new independent factor groups from the dimensions of EO and ESE while retaining the refined variables for PBP, attitudes, and AI adoption (see Table 3 and Figure 2). The results of the revised analysis revealed several key insights. Firstly, there was a positive correlation between the exogenous constructs: Proactiveness and Innovativeness (0.66), Seeking investors and PBP (0.29), and Opportunity development and PBP (0.32). Notably, Innovativeness, Opportunity development and PBP exhibited only an indirect effect on AI adoption through their relationships with other constructs. The strongest predictor of AI adoption was found to be positive attitudes toward AI, with a significant effect size of 0.53. Proactiveness showed a negative, though weaker, effect (−0.13) on AI adoption, while Seeking investors had a positive but weak effect (0.12). Both of these effects were statistically significant at p < 0.05.

Despite these findings, the proposed theoretical model could not be fully validated, as neither EO, ESE, nor PBP directly influenced attitudes toward AI. Instead, Innovativeness negatively impacted AI adoption when correlated with Proactiveness (EO). Conversely, PBP, in conjunction with Opportunity development and Seeking investors (ESE), exhibited a weak-positive direct influence on AI adoption. The dominant finding was that positive attitudes toward AI emerged as the most significant and independent predictor of AI adoption (see Figure 2).

In summary, the analysis revealed a complex interplay of factors where various dimensions of EO, ESE, and PBP both positively and negatively affect AI adoption. Despite the partial confirmation of hypotheses H1, H2, and H3, this research underscores that positive attitudes toward AI play a crucial role in driving AI adoption within the studied SMEs. Figure 3 (presented below) illustrates the final model of the research, showing the relationships between the studied variables in relation to the research (theoretical) model outlined in Figure 1.

5. Discussion

Our results offer new insights into AI adoption within hospitality SMEs amid the industry’s digitalization trend. Despite the potential benefits of AI for enhancing customer experiences and operational efficiencies, adoption rates among these businesses remain relatively low. This indicates that challenges persist in integrating AI technologies into operations.

EO and ESE are psycho-sociological constructs that emphasize internal attributes and beliefs driving entrepreneurial behavior [21]. EO includes traits such as risk-taking and innovativeness, while ESE focuses on an individual’s confidence in their entrepreneurial skills [18,64]. Based on the RQ posed at the beginning of this study, we hypothesized (see Figure 1) that managers with higher levels of innovation, proactivity, risk-taking, and confidence, and PBP would exhibit more positive attitudes toward AI and a greater propensity to adopt new technologies.

Our findings confirmed the significant role of positive attitudes in AI adoption, consistent with the established theoretical models (see Section 2.3). However, the direct relationship between these positive attitudes and actual AI adoption behavior requires further emphasis, as managers who hold favorable views of AI are more likely to engage with AI technologies actively. Accordingly, they may seek out relevant training, allocate resources for implementation, and foster a culture of innovation.

Notably, attitudes were not influenced by EO, ESE, or PBP (see Figure 2 and Figure 3). This surprising discrepancy suggests that our theoretical model does not fully capture the relationships between the investigated constructs, indicating the complexity of other factors influencing AI integration in hospitality SMEs. Based on our industry experience, we suspect that cultural and managerial factors may critically influence this discrepancy (particularly the personal values and the management style of the owner–manager). This is especially relevant, as many hospitality SMEs prioritize personalized customer service over technological innovation. Additionally, the omission of industry-specific factors such as labor intensiveness, a high proportion of a less-educated workforce, pronounced seasonality, and operational complexity could also impact these relationships. Furthermore, the high proportion of family-owned businesses in our study (61%) may have influenced these dynamics (see also Section 6).

Interestingly, our SEM analysis revealed that only one dimension of EO—Proactivity in conjunction with Innovativeness—has a modest, direct negative influence on AI adoption. A possible explanation for this might be that highly proactive and innovative family-owned hospitality SMEs perceive a negative impact from AI, viewing it as conflicting with their philosophy of offering personalized customer experiences. Conversely, one dimension of ESE—Seeking Investors in correlation with Opportunity development and PBP—exerts a modest, direct positive influence on AI adoption. This indicates that SMEs more inclined to seek investors and develop opportunities while maintaining a positive view of their business performance are more likely to create an environment conducive to AI adoption. However, this inclination does not directly influence managers’ personal attitudes toward AI. These findings depart from previous research, which often reports a positive association between EO dimensions and digital transformation in (non-hospitality) SMEs [74,75,76]. It seems essential to further investigate how these variables interact to shape behaviors toward technology adoption.

Overall, our results suggest that EO, ESE, and PBP do not influence managers’ attitudes toward AI and have only a limited, direct impact on AI adoption. While validated research instruments were used (see Section 3.1), it appears our current conceptual approaches might need refinement to understand technology adoption in hospitality SMEs better. This aligns with Mogaji et al. [47], who call for more nuanced conceptual frameworks in AI research.

The finding that EO, ESE, and PBP do not influence attitudes suggests that other factors may be more influential in shaping positive attitudes toward AI. The model confirmed 31% of the total variance in AI adoption, indicating additional variables and/or contextual factors need exploration. Future research should identify these hidden factors and refine theoretical models to capture the complexities of AI adoption in hospitality SMEs.

However, our findings highlight the need for further investigation into EO and ESE’s impact on hospitality SMEs. This need arises from the contrast between our research results and those observed in other sectors. Namely, studies in the metal and machinery sector in Indonesia [77] and among family businesses in various economic sectors in India [18] have demonstrated that EO significantly influences the digitalization process. Similarly, Cvijić Čović et al. [64] emphasize EO’s role in SME digitalization across different sectors in Serbia, although their study did not specifically address hospitality businesses. These findings suggest that while EO might be pivotal for digitalization in various sectors, its role in hospitality SMEs requires focused research. The unique operational characteristics and market dynamics of the hospitality industry may influence how EO and ESE impact technology adoption.

The discrepancy between our findings and those from previous research could also be attributed to the specific managerial characteristics inherent to the hospitality industry. Hospitality managers might intentionally reject AI as part of their customer-centric strategy, perceiving that embracing AI could undermine their efforts to differentiate from competitors adopting similar technologies. However, AI does not necessarily disrupt the provider–customer relationship. Instead, it has the potential to tailor customer experiences (e.g., Customer Relationship Management—CRM) and optimize business processes, such as improving efficiency, streamlining internal processes, and enhancing data-driven decision-making [35]. This is also important, as the study results indicate that managers are relatively dissatisfied with their SMEs’ profitability (P6; M = 3.79).

Given these factors, further studies are essential to understand the adoption of AI in the hospitality sector. Future research should also explore managerial motivations, industry-specific dynamics, and how AI can be integrated without compromising traditional and personalized aspects of hospitality. This will provide a more comprehensive perspective on AI adoption barriers and facilitators in hospitality SMEs.

Our study has also highlighted relatively low positive GAAIS scores averaging M = 2.70 (SD = 0.07) and negative GAAIS averaging M = 2.97 (SD = 0.16). These scores are notably lower than those that Schepman and Rodway [18] reported in the UK. Our findings, along with low AI adoption rates (see Table 3 and Appendix A), underscore the need for exploring barriers to AI integration in hospitality SMEs. In this view, the demographic profile of managers in our study (many with secondary education and aged 46–55, with 70% being SME owners) suggests a potential gap in new technology adoption among experienced managers. This underscores the importance of developing strategies to help managers overcome obstacles to digital transformation, as previously underscored by Kaya et al. [17].

Finally, our study’s findings suggest that positive attitudes are the primary drivers of AI adoption, while negative attitudes have no impact on it. This underscores the importance of promoting a favorable view of AI rather than counteracting skepticism.

6. Conclusions

Our study highlights the complexities of AI adoption within hospitality SMEs. Despite clear digitalization trends and the potential benefits of AI for improving customer experiences and operational efficiencies, entrepreneurial characteristics alone do not significantly drive AI adoption in this sector. Contrary to theoretical expectations, both EO and ESE did not significantly influence attitudes toward AI. Instead, the most crucial factor for AI adoption is the positive attitudes of managers toward AI (see Figure 2 and Figure 3).

However, the sample size may limit the generalizability of these findings, and the use of convenience sampling may have impacted the sample’s representativeness, suggesting a need for broader and more diverse samples in future research. While entrepreneurial traits are theoretically important, their practical impact on the adoption of AI in the hospitality sector is limited. This emphasizes the necessity to focus on how managers personally perceive and engage with new technologies rather than relying (solely) on their entrepreneurial traits. Industry-specific factors may additionally contribute to the reluctance to adopt AI, especially in contrast to other sectors where entrepreneurial characteristics have positively influenced digitalization efforts.

To address these challenges, implementing targeted educational interventions aimed at enhancing managerial (positive) attitudes toward AI is essential. Given the current low adoption rates and the complexities involved in integrating new technologies, further research should explore industry-specific factors such as guest orientation and resistance to change, which may impact AI integration in hospitality SMEs. This need aligns with the EU Commission’s efforts to advance AI skills among SMEs.

Theoretically, this research contributes to the understanding of AI and entrepreneurship by examining the interplay among various organizational factors and AI adoption in hospitality SMEs. First, by focusing on hospitality SMEs, this study specifically examines AI adoption within this sector, a context that has been relatively underexplored in AI research. This focus is particularly relevant given the significant role SMEs play in the EU economy. Second, the research highlights the importance of entrepreneurial traits in influencing managers’ attitudes towards AI adoption. This adds a new dimension to the understanding of AI adoption by linking it to psycho-sociological and behavioral factors that drive decision-making in SMEs. Third, by utilizing the GAAIS scale, which has been specifically designed for measuring attitudes towards AI, and SEM, this study provides empirical evidence that supports the theoretical relationship between managerial attitudes and AI adoption, offering a more robust understanding of the importance of positive attitudes in the AI adoption process. Fourth, this study builds on the TAM by incorporating additional variables that address the unique challenges of AI adoption. While TAM traditionally focuses on PEOU and PU, this research examines how EO, ESE, and PBP influence attitudes and AI adoption. This extension allows for a more comprehensive understanding of the factors influencing AI adoption in the hospitality sector, considering that research results indicate that entrepreneurial traits have no influence on attitudes, though they directly influence AI adoption. This study reaffirms TAM’s emphasis on attitudes as a critical factor and suggests that AI in this sector is still in its early stages of adoption. By integrating new variables and focusing on the specific context of hospitality, this study provides a richer framework for understanding AI adoption in this sector.

Ultimately, for practical application, SMEs should enhance their awareness of technological advancements and evolving market dynamics to adopt AI more effectively. Policymakers and industry stakeholders, in cooperation with academia, should collaboratively address industry-specific gaps in AI knowledge and create a supportive ecosystem to facilitate AI adoption.

Future research should consider qualitative methods for deeper insights and investigate external and organizational factors influencing AI adoption. Longitudinal studies could determine whether the limited impact of entrepreneurial traits is related to the current low adoption rates (and vice versa). Additionally, exploring how tradition, particularly in family-owned businesses, and different perspectives on AI acceptability could offer a more comprehensive understanding of the benefits and challenges of AI adoption in hospitality SMEs.