The advancement of technology has made the applications of artificial intelligence (AI) more prevalent. Researches into the applications of AI in business environments have taken employees’ intention to adopt artificial intelligence (AI) applications and their big data analytical capability (BDAC) into consideration, focusing on the opportunities and applications of AI (von Joerg and Carlos, 2022). AI is an umbrella term that refers to a range of complementary techniques that have been developed from statistics, computer science, and cognitive psychology (Procter et al., 2020, p.5). Digitalisation and greening of economy are the twin concepts to promote sustainable development. One example of new digital technologies is AI, which can collect, assess, and analyse data, as well as to communicate the results to a wider range of applications (Carayannis & Morawska-Jancelewicz, 2022).

AI enables the development of BDAC and is leading the Industry 5.0 revolution (Carayannis & Morawska-Jancelewicz, 2022; Tirgil & Fındık, 2022). AI can be used to analyse large amounts of data, identify relevant data, utilise the data to improve operations, deliver actionable insights, and create novel economic and business values (Giovanni, 2021). Adopting AI applications can reduce operating costs and increase the value of a company’s brand, thereby enhancing the company’s performance (Silva et al., 2021). Employees’ intention to adopt AI applications is important as their actions could influence the competitiveness of the company. The deployment of AI applications in business might shift the business activities towards its use to create values in new ways for the company. Employees that intend to learn are able to gather information about the market and tend to be very flexible as they need to react to various changes in the business environment. With those employees that are willing to learn, the company could build on their existing knowledge and modify internal practices to improve product performance (Otioma, 2022).

Over the past decade, the global economy has shifted from industry-oriented to digital economy-oriented (Harris et al., 2021). Moreover, rapid globalisation and the digital economy have created a highly uncertain, volatile, and intensely competitive business environment. To sustain development, survive in the market, and maintain competitiveness, companies must prioritise the development of dynamic capabilities (Li & Liu 2014). In addition, companies are increasingly investing in information technology (IT) to achieve real-time information sharing and streamlining operations to enhance their competitiveness (Agrawal et al., 2021). The real-time information generated using IT contains substantial amounts of data on transactions with supply chain partners. IT resources alone do not create value unless data are utilised to integrate internal and external stakeholders (Yu et al., 2021). Therefore, to benefit from big data, companies must have flexible IT resources to integrate their relationships with internal and external partners. Fishbein and Azjen (1975) conceptualised human behaviour as reasoned action that follows from behavioural intention. The theory of planned behaviour (TPB), which was developed from the theory of reasoned action (Ajzen & Fishbein, 1980), suggests that the behavioural attitude, subjective norms, and perceived behavioural control can determine a person’s behavioural intentions (Ajzen, 1991).

The different facets of AI present a range of business opportunities, but at the same time, also elicit complex regulatory challenges. The applications of AI on customer service, data quality management, data usage experience, and acquisition intention of big data analytics (Kwon et al., 2014), and digital humanities (Dukhanov et al., 2016) have demonstrated the pre-disciplinary demands on employees’ intention to adopt AI applications and BDAC. In an environment dominated by big data, traditional analytical methods cannot effectively extract critical business information. Therefore, companies are focusing on acquiring BDAC, which can dynamically assist managers in the reallocation of corporate resources. BDAC is a holistic process to manage, process and analyze 5 Vs (volume, variety, velocity, veracity, and value) to create actionable insights for sustained competitive advantages (Sun & Liu, 2021). The application of big data in supply chain activities is crucial for generating unique corporate values and increasing companies’ competitive advantages (Wang et al., 2020). BDAC can enhance sustainable corporate development in the domains of the Internet of Things, social networks, and supply chains. Studies have revealed that companies with BDAC can understand market conditions more broadly than their competitors because BDAC enables these companies to identify valuable information from big data, identify customer needs, improve decision-making, and develop superior products due to improved organisational performance.

BDAC has become increasingly critical in all aspects of business management. This phenomenon has been driven by the existence of large-scale data and managers’ desire to root their decisions in data. Relevant research has demonstrated that supply chain and operational management functions are among companies’ most critical sources and uses of data. Therefore, companies’ decision-making processes benefit from an increased use of big data analytics. However, a lack of understanding persists regarding what determines a company’s ability to develop BDAC for gaining a competitive advantage and improving operational performance (Jha et al., 2020), as well as achieving supply chain integration and operational flexibility. Accordingly, a substantial gap remains in the literature regarding the correlations of BDAC with integration capability, team collaboration, and corporate performance (Jha et al., 2020; Pan & Guo, 2021). Understanding how BDAC affects integration capability and team collaboration can enable companies to enhance their global competitiveness. Although high-tech firms are partly affected by the COVID-19 pandemic, as seen in Fig. 1, high-tech firms in Taiwan have maintained their production outputs and remained the stable growth. As Taiwan has a high rate of BDAC application, this study selected Taiwanese high-tech firms to be the research objects. The resource-based theory and the contingency approach were utilised to examine the relationships of the employees’ intention to adopt AI applications, firms’ BDAC, firms’ integration capability, and team collaboration with operational performance.

Fig. 1
figure 1

High-tech product value in Taiwan. Source: MOEA (2022)

Employees’ intention to adopt AI applications implies an implicit behaviour, and BDAC refers to the skills required for using the AI applications. Both variables are the critical factors to affect the consequential behaviour pattern and, in turn, firms’ operational performance (Pika et al., 2021; Wu et al., 2020). There are very few studies focusing on the role of employees’ intention to adopt AI and BDAC in high-tech firms. The research purposes are as follows:

  1. (1)

    To explore how employees’ intention is related to integration capability and team collaboration in AI application processes.

  2. (2)

    To explore how big data analytical capability is related to integration capability and team collaboration in AI application processes.

  3. (3)

    To elucidate the relationship between integration capability and team collaboration in AI application processes.

  4. (4)

    To understand the effect of integration capability and team collaboration on operational performance.

The findings of this study can fill the research gap in exploring the roles of employees’ intention to adopt AI applications and BDAC in integration capability and team collaboration, as well as their effects on operational performance. Moreover, this study also attempted to determine the mediating effect that influences operational performance of high-tech firms.

Literature Review

In the digital economy, information flow in the supply chain is crucial to a company’s survival and sustainable development. Faced with continuous technological advancements in the last decades, businesses are seeking new ways to adapt their business models to a digital and connected environment (Veile et al., 2022). In this interconnected ecosystem, strategies in digitalisation and AI applications have become cornerstones in new product and service development to enhance competitiveness (Palmié et al., 2022). Big data, which has attracted the attention of numerous companies, comprises a large amount of the structured and unstructured data collected by companies (Shamim et al., 2020). As BDAC can provide useful information for decision-making, it has become increasingly crucial for performance improvement. Therefore, a company must cultivate a unique BDAC that is difficult to imitate (Gupta and George 2016). However, because of the high volume, velocity, and variety of big data, extracting useful information from it by using conventional analytical methods is difficult. BDAC allows for the ability to process, visualise, and analyse substantial amounts of structured and unstructured data. Moreover, it enables unique insights to be obtained for data-driven operational planning, decision-making, and execution (Monino, 2021; Srinivasan & Swink, 2018). BDAC enables companies to integrate and analyse data to realign business processes in a rapidly changing environment. BDAC has been used in corporate procurement, product design, and R&D, and it positively affects business performance (Dubey et al., 2019).

Artificial Intelligence (AI)

This study is based on the theoretical framework of the unified theory of acceptance and use of technology (UTAUT) (Marikyan & Papagiannidis, 2021; Venkatesh et al., 2002), with some modifications to fit the research aims and data. By better understanding the predictors associated with adopting AI and related collaboration technologies, high-tech firms can achieve optimised operational performance.

The definition of AI has been debated in both the academia and industry, and little consensus has been reached on the terminologies used in different fields of AI applications. AI is a general term that describes the use of computers to simulate intelligent behaviour with minimal human intervention (Liu et al., 2022). AI is based on the idea of allowing computers and machines to mimic human behaviours and ultimately “think” like humans. These machines are meant to perform simple tasks and make decisions based on data they have gathered (Andrews et al., 2021). Furthermore, AI can be divided into artificial general intelligence (AGI), artificial narrow intelligence (ANI), and artificial super intelligence (ASI), as shown in Fig. 2 (Kelly et al., 2023).

Fig. 2
figure 2

Hierarchy of AI. Source: Kelly et al. (2023)

  1. (1).

    ANI includes modern AI systems, such as voice recognition software (e.g., Apple’s Siri), which assists users via machine learning and cannot transfer knowledge across systems or tasks.

  2. (2).

    AGI is currently at the theoretical stage and will be able to achieve goals autonomously and transfer leaning results within a wide range of scenarios. Such abilities can enable AGI agents to possess intelligence far beyond human capability, and may lead to developments in complex issues, such as human health and global warming.

  3. (3).

    Finally, ASI involves agents that will function at a higher level of intelligence than human beings’ capabilities. ASI is the most accurate form of AI, as it will be capable of pioneering discoveries in general, scientific, academic, creative, and social fields, potentially leading to the redundancy of human beings.

Digitalisation, Big Data Applications, and Innovation

The acquisition of information on customers and their behavioural actions in digital and AI environments can allow firms to pay increasing attention to user big data and analysis (Ribeiro-Navarrete et al., 2021). Figure 3 shows the milestones of industry innovations (Carayannis & Morawska-Jancelewicz, 2022; Tirgil & Fındık, 2022).

Fig. 3
figure 3

Industry innovations milestones

Figure 4 illustrates the digitalisation process, which demonstrates the advances in digitalisation, and the move toward increasingly flexible manufacturing processes. The digitalisation process has created new opportunities and choices for industrial companies across the globe. Digitalisation infrastructure and BDCA intelligently connect energy systems, buildings, and industries to improve our daily lives and work through greater efficiency and sustainability. AI, by integrating software, virtual applications, hardware, products, systems, and solutions, makes infrastructures more intelligent, responsive and responsible.

Fig. 4
figure 4

Digitalisation trend

The applications of AI in industry innovations are fast growing. In recent years, Industry 5.0 represents the convergence with quadruple/quintuple helix models of innovations (Q2HM), and can be considered as the answer to the demand of a renewed human-centred/human-centric industrial paradigm, from the (structural, organisational, managerial, knowledge-based, philosophical, and cultural) reorganisation of the production processes to the generation of positive implications within the business perspectives, as well as all the components belonging to the innovation ecosystem (Carayannis et al., 2021, p. 5; Carayannis, 2021a, b, c, d). Industry 5.0 relies on three core elements: human-centricity, sustainability and resilience. The Q2HM models play an important role in fostering the shift from technical to social innovations, especially towards a more balanced human-centric and techno-centric approach (Table 1).

Table 1 Evolution of techno- and human-centric innovation

Figure 5 shows the annual labour force in Taiwan from 2020 to 2021. As seen, the labour force has slightly decreased, partially affected by the lockdown policies due to the COVID-19 pandemic. Contrary, in 2021, the revenue of electronic products sales is about NT$ 4,513 trillion, approximately equal to US$161 trillion. It is a 20% increase compared to NT$3,748 trillion in 2020, as shown in Fig. 6. This increase was contributed by the AI applications in Taiwan high-tech firms. Take the light-emitting diode (LED) industry for example, the technological development has significantly increased the luminous flux and light intensity. Since 2000, the applications of LED have achieved a great growth. Other examples of AI applications include the spectrophotometric process and surface-mounted technology (SMT), as the manual works have been greatly replaced by AI. As the demand of manpower has significantly shifted from human to AI, in such technology-led manufacturing process, employees’ intention to adopt AI applications is crucial to the integration capability and team collaboration. If the employees have low intention to adopt AI or are not training properly, some equipment may not operate independently, especially for some implicit technology. Numerous industry reports and consultancy white papers have revealed that BDAC is conducive to companies. Renowned corporations such as Apple, Dell, and Samsung are actively striving to improve their supply chain processes and create new business opportunities through their powerful BDAC. Companies that intend to obtain profound insights from big data need to consider the challenges of introducing the technology, especially the investment of resources, because overcoming these challenges is an essential element of developing a unique BDAC (Gupta & George, 2016). Resource investment also enables employees to improve their digital knowledge and big data skills (Dubey et al., 2019).

Fig. 5
figure 5

Labour force (thousands). Source: MOEA (2022)

Fig. 6
figure 6

Electronic products sales revenue (thousand/NT$)

Team Collaboration and Integration Capability

In an intensely competitive business environment, companies must establish close cooperative relationships with upstream and downstream supply chain partners. Thus, in this industrial environment, supply chain integration is a critical competitive strategy (Jha et al., 2020) in which companies collaborate with their partners and manage intraorganisational and interorganisational information, products, and processes across the supply chain from upstream suppliers to downstream customers (Naimi-Sadigh et al., 2021). The purpose of this integration is to maximise customer value (Huo et al., 2014). According to Irfan and Wang (2019) and Liu et al. (2018), supply chain integration is divided into two key dimensions: integration capability and team collaboration. Integration refers to the degree to which a company structures its intraorganisational resources, practices, and procedures into collaborative, synchronised, and manageable processes and systems across functions to fulfil customer requirements. Through the reallocation of resources and tasks, even minor changes in integration can considerably influence a company’s competitiveness (Wang & Wang, 2019). By strengthening its team collaboration, a company can provide high-quality services to customers and achieve excellent performance. Integration capability refers to the degree to which a company collaborates with its supply chain partners (including upstream suppliers and downstream customers) to support its development in the dynamic business environment (Irfan & Wang, 2019). With the emphasis on coordination with external parties and resources, integration capability between a company and its supply chain partners demonstrates their ability to agree to invest jointly in resources, share information, and assume responsibility for achieving operational goals (Zhao et al., 2011).

Integration capability helps a company to differentiate itself from competitors and arises from critical management decisions (De Luca & Atuahene-Gima, 2007); it creates a harmonious working environment that favours employees’ intentions to collaborate. Helfat and Winter (2011) stated that integration capability may be dynamic or operational, depending on the nature of the capability and the purpose of its use. Bhaskaran and Krishnan (2009) observed that team collaboration alone does not ensure the performance of new product development. The effects of product development on marketing, R&D, manufacturing, suppliers, and customers are determined by integration capability (Fain et al., 2011). Successful integration capability allows a company to engage in production at reduced costs and enables a superior customer value proposition (Chang et al., 2016).

Moving from Traditional Operation Strategies to AI Business Operation

As AI technologies have led to the creation and adaptation of new business models, businesses have adapted strategies to the new ecosystem. While traditional business operations focused their objectives and strategies on existing methods, the new business operation prioritised circular supply chains and big data analysis to improve decision-making. Business operational strategies focused on the newly established digitalisation environments, such as AI and digital platforms, which are common in high-tech firms. In this novel paradigm, AI applications and collaboration-centred strategies have become crucial options for companies to optimise their operational performance.

Digitalisation technologies and AI applications are expected to accelerate the pace of innovation and enhance firms’ innovation performance (Kastelli et al., 2022). However, few studies have investigated whether this expectation is bound by other conditions in adopting new digital-based systems, such as integrating internal and external resources and up-to-date technologies, which are capable to optimise business operational performance. In this context, employees’ intention to adopt AI applications play crucial role in affecting the business operational performance.

Hypotheses Development

In theory, as well as in practice, the definition of behavioural intention has been widely discussed from different perspectives in the academia and industries. The TPB argues that behavioural attitude, subjective norms, and perceived behavioural control can determine a person’s behavioural intentions (Ajzen, 1991). In addition, compared to the rational behaviour theory, the TPB has been used in more research applications. Extant research argues that the application of TPB is a very rigorous, consistent, highly universal, and sound theoretical framework (Jun & Arendt, 2016). In practice, Zhang and Prybutok (2005) argued that the measurement of behavioural intentions of e-services may be observed from three perspectives, namely, I intend to use e-services; I plan to use e-services frequently; and I plan to use e-services whenever I need electronic services in the future. Several literature sources claim that a consumer’s satisfaction and emotional reaction are positively correlated with their behavioural intention, which includes customer loyalty, recommendations to other potential customers, and word-of-mouth, as well as a willingness to pay a higher price (Jani & Han, 2015). The formation of behavioural intentions includes the following components: cognitive factors, emotional factors, behavioural factors, and loyalty. Furthermore, several notable findings have been reported, including the empirical verification that service quality, service value, and satisfaction may all be directly related to behavioural intentions (Cronin et al., 2000).

Mazzucato argued that openness and collaboration are not a nice complement, but rather a critical factor for success (Mazzucato, 2018, p.5). Adopting a resource-based perspective of product development, Ahmadi et al. (2014) claimed that realising the benefits of collaboration practices requires the development of integration capability. Newell et al. (2008) observed that biomedical innovation involves intense collaboration across disciplines, occupations, and organisations. A nation’s ability to integrate basic science and clinical development, and its ability to collaborate with diverse organisations, has been identified as critical to improving its performance. These arguments reveal the correlation between integration capability and team collaboration. Teams in a company have their own perspective of integration-related activities. The perspectives of design, marketing, and manufacturing personnel within a company are typically determined by the tasks or expertise of those personnel, making cooperation difficult and frequently causing conflict and confusion (Truffer et al., 2001). This fact demonstrates the importance of integration to managing, coordinating, and efficiently performing various activities. Moreover, in terms of functional domains, collaboration may not significantly affect, or may even negatively affect, performance if disagreements among team members concerning product development are not properly addressed. Accordingly, collaboration alone does not guarantee the performance of product development unless integration also occurs. A team with high integration capability is likely to exhibit great collaboration among team members.

The more aggressive an individual’s attitude is toward behaviour, the higher the behavioural intention. Nevertheless, when an individual’s attitude toward behaviour is more negative, the behavioural intention is lower (Ajzen & Fishbein, 1980). Based on the above-mentioned literature, this study adopted the TPB to explore the relationship between employees’ intention, integration capability, and team collaboration. The following hypotheses are posited regarding the effect of employees’ intention to adopt AI applications.

  • Hypothesis 1 a (H1 a ): Employees’ intention to adopt AI applications is positively correlated with integration capability.

  • Hypothesis 1 b (H1 b ): Employees’ intention to adopt AI applications is positively correlated with team collaboration.

In the digital economy, employee retention and training are primary competitive strategies. Companies require experienced employees with extensive knowledge and skills to create, integrate, utilise, and reconfigure BDAC (Wang et al., 2020). These employees also need to update corporate resources and core competencies, strengthen their knowledge of the internal digital economy, and acquire relevant external knowledge for further enrichment. Knowledge accumulation within companies is a core foundation of technological innovation (Fonseca et al., 2019).

A company represents a bundle of resources or capabilities. Among these resources, IT has a major influence on a company’s ability to develop unique capabilities (Shamim et al., 2020). For example, advanced IT systems such as AI, cloud computing, and supply chain communication systems are characterised by rapid data and information processing for restructuring and optimising activities in the supply chain. Specifically, companies use big data to gain new insights, discover new business opportunities, understand product and process design, and ascertain supplier and customer needs (Yu et al., 2021). According to this view of dynamic capabilities, BDAC can enhance a company’s information processing capabilities and thereby create value for the company.

The appropriate application of IT can enable the efficient and automatic transmission of supply chain information that is typically scattered, such as product availability, inventory levels, shipment status, and production requirements (Wang & Wang, 2019). The early stage of industry digitisation was characterised by serious data consistency problems in large database systems and high error rates in decentralised systems, such as supply chains. By contrast, currently, the large amounts of unstructured and real-time supply chain data collected by companies can be interpreted effectively and quickly through big data analysis. Accurate, reliable, and consistent information can be transmitted to corporate procurement, production, marketing, and R&D departments (Rezaee & Wang, 2018). The integration of internal company information can improve each department’s efficiency and facilitate communication and cooperation among the departments (Ganbold et al., 2020). Companies with excellent big data processing capabilities can continually monitor the changing business environment to promote team collaboration.

The capability to implement big data effectively in the supply chain and the information obtained from big data enable companies to understand market needs deeply, generate new and unique insights, and establish a cooperative relationship with their supply chain partners. The application of big data can also help companies deliver consistent and high-quality information to external partners (Rezaee & Wang, 2018). Shared information enables companies to interact with all parties in the supply chain to achieve close coordination and business linkages. This interaction helps companies to monitor supply chain processes, thereby enhancing production, demand and supply planning, and operational control. Cai et al. (2016) indicated that the dynamic capabilities of companies to acquire, deploy, assemble, and reconfigure IT are positively correlated with rapid responses to changes in the business environment and with collaboration between organisations (Jha et al., 2020).

Based on the above, the following hypotheses BDAC on integration capability and team collaboration are proposed:

  • Hypothesis 2 a (H2 a ): BDAC is positively correlated with integration capability in AI application processes.

  • Hypothesis 2 b (H2 b ): BDAC is positively correlated with team collaboration in AI application processes.

AI is characterised by intense competition, a high degree of uncertainty, and rapid change. Therefore, companies must partner with their external suppliers to develop unique resources and capabilities for maintaining their competitive advantages. However, companies are typically reluctant to integrate externally because the information exchanged with external suppliers is often asymmetric unless they have achieved adequate team collaboration. Internally integrated companies possess adequate communication, information sharing, and cross-departmental cooperative capabilities, so that they can identify the key challenges related to their suppliers and customers. This identification is conducive to the maintenance of beneficial strategic partnerships (Zhao et al., 2011). It enables companies to align with external partners in terms of information, processes, technology, and operational measures. In addition, the aforementioned identification can enable companies to reconfigure their internal operating processes and production systems, as well as manage external uncertainties and related challenges jointly. Several studies have suggested that integration capability is an antecedent of team collaboration (Chen et al., 2015; Jayaram et al., 2011).

Numerous arguments have been submitted regarding the relationship and prioritisation of integration capability and team collaboration. For example, Zhao et al. (2011) suggested that integration capability enables the establishment of team collaboration because organisations must first develop integration capabilities through systems, data, and process integration before they can engage in meaningful team collaboration. Horn et al. (2014) proposed that integration capability is a precondition of team collaboration with suppliers that has a strong positive influence on successful global sourcing. Titah et al. (2016) argued that full integration results in optimal firm profitability, inventory management, and customer service when the supplier and retailer have shared objectives. Furthermore, Graham (2018) indicated that internally based environmental protection efforts are essential antecedents to the development of internal environmental protection capabilities, which may be useful in the extension of environmental protection efforts to the supply chain level.

In integration capability, the importance of establishing close interactive relationships with customers and suppliers is recognised. In team collaboration, a company’s individual departments and functions are recognised to function as components of an integrated process. Both perspectives are crucial for enabling supply chain partners to act in concert to maximise the value of the overall supply chain (Flynn et al., 2010). Integration involves not only the integration of external resources, but also the effective integration of external resources and core competences with the resources of different departments in a company (Giovanni, 2021). Integration capability and team collaboration play different roles in a company. Hence, to analyse the effect between Integration capability and team collaboration in the high-tech firms, it is hypothesised:

  • Hypothesis 3 (H3): Integration capability is positively correlated with team collaboration in AI application processes.

Corporate integration and cross-functional strategic coordination and integration with supply chain partners are key factors for improving the operational performance of companies (Ganbold et al., 2020). Integrating with external supply chain partners to establish long-term strategic partnerships enables companies to identify challenges and communicate rapidly for completing corresponding design and production tasks. Such integration enables companies to reduce their product distribution and delivery times, which allows them to improve efficiency and rapidly introduce new products to markets. They can also solve the problem of excessive costs caused by inventory obsolescence. Finally, companies that have implemented integration capability can produce and provide higher-quality products to supply chain partners more flexibly and generate higher operational performance than can companies that have not implemented integration capability (Yu et al., 2021). Meanwhile, scholars have indicated that both the coordinated integration of knowledge and systemic integration of knowledge have positive effects on high-tech new venture performance and product innovation (Guo et al., 2019).

Team collaboration reduces functional barriers and engenders cross-functional cooperation to ensure customer satisfaction. In contrast to the functional silos associated with traditional departmentalisation and specialisation, team collaboration is expected to affect operational performance positively. BDAC promotes team collaboration and thus disrupts the independent operations of functional departments.

BDAC enables employees to cooperate and helps the management to adapt quickly to changing market demands. Irfan and Wang (2019) suggested that production efficiency can be improved by increasing the flow rate of raw materials and information in the integration process. In addition, numerous studies have confirmed that team collaboration could improve cost, quality, delivery, flexibility, and process efficiency. Other researchers have demonstrated that team collaboration is positively correlated with operational performance (Flynn et al., 2010).

Although some researchers did not confirm any direct relationship between integration and operational performance (Gimenez & Ventura, 2005), other studies identified a positive relationship between integration and operational performance, including operational performance in new product development (Silva et al., 2021). On the basis of the aforementioned discussion, the following hypotheses are proposed:

  • Hypothesis 4 a (H4 a ): Integration capability is positively correlated with operational performance in AI application processes.

  • Hypothesis 4 b (H4 b ): Team collaboration is positively correlated with operational performance in AI application processes.

Research Methodology

Research Design

This study selected high-tech firms, which were defined as firms that are officially registered with the government and are expected to have a high potential in innovation activities and innovation production (Yu et al., 2021), as the research objects. Considering are currently the leading manufacturers of semiconductor, electronic/electric, and eco-labelled products for AI applications in the manufacturing process (TRPC, 2020), this study focused on high-tech firms in Taiwan.

Confirmatory factor analysis (CFA), structural equation modelling (SEM), and related statistical analyses were used for statistical analysis. The measurement model was evaluated on a Likert 5-point scale and analysed using SPSS 26 and LISREL 10.20.

Data Collection

Data were collected for March 2021 to August 2021.The contact information of the high-tech firms was obtained from the commercial data information provider Dun and Bradstreet (Table 2).

Table 2 Characteristics of firms

Survey Design and Analytical Procedure

The questionnaire and an explanatory letter were emailed, mailed, or hand delivered to employees of the selected firms. No more than one questionnaire was sent to any single department. To test the possible representativeness of the participating firms and the nonresponse bias, a MANOVA analysis was conducted for comparing early and late respondents in terms of all the variables. The results revealed no significant difference between the early and late respondents at p < .05 (Novak, 1995). The sample of firms that responded to the questionnaire was compared with the qualifying group that did not respond in terms of age of firms and products/applications of the firms, all of which were found to be almost identical between both groups. Therefore, the sample of firms analysed was assumed to be representative of the target population.

The First Stage of the Questionnaire Survey

The first stage of the questionnaire survey collected basic information on the firm (12 items) and operational performance (7 items). Since the study was intended to evaluate operational performance, this survey was separated from those of other variables. Of the 600 questionnaires distributed, 349 were recovered. Among those recovered, forty-one questionnaires were invalid, and 308 were effective.

The Second Stage of the Questionnaire Survey

In the second stage, the questionnaire collected data on employees’ intention to adopt AI applications (5 items), BDAC (5 items), integration capability (5 items), and team collaboration (9 items). A total of 308 questionnaires were distributed to the respondents who returned questionnaires in the first stage. Of the 278 questionnaires recovered, eight invalid samples were eliminated, totalling 269 valid samples. This study used program level as a basis for evaluating operational performance and analysed the 269 returned questionnaires.

Sample Characteristics and CMV

Common method variance (CMV) depends on respondents’ affective states and their tendency to provide socially desirable responses (Chang et al., 2010). In general, three items of remedies exist for assessing CMV: (1) procedural remedies, which are performed in the ex-ante phase of research; (2) statistical remedies, which are performed in the ex-post phase of research (Chang et al., 2010); and (3) remedies based on a reversed item, namely, the time-separation test and Harman’s one-factor test. The time-separation test was conducted on data that were collected in two stages, namely, 4-week and 6-week intervals. In this study, the 4-week to 6-week lag appeared to be insignificant because the questionnaire was administered from March 2021 to August 2021, and in most cases, the separation interval was approximately 3 to 4 months. Thus, the time lag did not create a bias in the results. To test for CMV, this study evaluated a correlated uniqueness model (Podsakoff et al., 2003). The model accounts for method effects by allowing measurement of error terms of constructs using the same method correlated in the measurement model. Comparing the correlated uniqueness model with the original measurement model identified no significant change in model fit or in any of the loading parameters, indicating that common method biases did not significantly affect data analysis results.

With respect to ex-post statistical remedies for assessing CMV, Harman’s one-factor test was conducted to determine whether common method bias influenced the data (Chang et al., 2010). If a single extracted factor explains a majority of the variance of the data, then CMV exists. In this study, the exploratory factor analysis was employed on all variables simultaneously, and the unrotated factor structure was not a single-factor structure. Specifically, the independent variables and the dependent variable did not load onto the same factor. Moreover, the first factor explained 18.88% of the variance in the data. Therefore, according to the Harman test, CMV was not a concern in this study. The two-step procedure, which assesses the reliability and validity of measures before their use in the full model (Anderson & Gerbing, 1988), was used to examine the measurement model and subsequently the structural model.

Reliability and Validity

CFA was performed to test the measurement model and assess the construct validity. The five latent variables of the CFA model were allowed to co-vary freely. The parameters of all the models were estimated using the maximum likelihood method by taking the item covariance matrix as the input. CFA results show that the measurement model fit the collected data χ2/df = 1.276, GFI = 0.93, NFI = 0.95, CFI = 0.96, and SRMR = 0.055 RMSEA was 0.021 for the model with a 90% confidence interval of 0.017–0.026. All the model-fitted indices exceeded commonly accepted levels that have been published in the literature (Kenny, 2015); thus, the measurement model had a satisfactory fit with the collected data.

Table 3 presents the measurements, items, response formats, composite reliability (CR) values, and factor loadings of the measures. The CR values were used to evaluate the internal consistency of each model construct. The results show that all the model constructs had CR values that exceed the recommended cut-off of 0.70 (Kenny et al., 2014). In addition, all the construct loadings exceeded 0.7, suggesting that the indicators used to measure each construct were related to the respective construct.

Table 3 Analysis of measurement model

The average variance extracted (AVE) was calculated for each construct. The results show that all the AVEs exceeded the recommended cut-off of 0.5 (Sivo et al., 2006). Thus, the model constructs satisfied the criteria for convergent validity. Discriminant validity is the degree to which two similar but conceptually distinct measures differ and can be tested by comparing the square root of the AVE of each construct with the corresponding inter-construct correlation coefficients. As presented in Table 4, for each construct, the square root of the AVE exceeded the corresponding inter-construct correlation, indicating the presence of discriminant validity. Accordingly, the constructs in our model measured theoretically distinct concepts.

Table 4 Discriminated validity: correlations and AVE

Description of Variables

The questionnaire was first prepared in English and then translated into Chinese (simplified and traditional). The Chinese version was subsequently back-translated by a third party to ensure accuracy (Chen, 2015; Chan, 1998). The three translations indicated no substantial differences in the meanings of the scales. This study used a Likert scale going from totally disagree (1) to totally agree (5).

Intention to adopt AI applications (IAAI) was measured using a five-item scale adopted and revised from Chai et al. (2021): IAAI-1, it is very convenient to use the products and services that use the latest AI technologies; IAAI-2, I feel confident that AI applications will follow the instructions I give; IAAI-3, AI technology gives me more control over their daily lives; IAAI-4, I prefer to use the most advanced AI technology available; and IAAI-5, I like AI technology that allows me to tailor things to fit my own needs

Big data analytical capability (BDAC) enables firms to increase their information processing capacity, according to which they collect and analyse data from various sources to obtain insights for their operations. This variable was measured using a five-item scale adopted and revised from Srinivasan and Swink (2018): BDAC-1, I use advanced analytical techniques (e.g., simulation, optimisation, regression) to improve decision-making; BDAC-2, I easily combine and integrate information from many data sources for use in our decision-making; BDAC-3, I routinely use data visualisation techniques (e.g., dashboards) to assist users or decision-maker in understanding complex information; BDAC-4, Our dashboards display information, which is useful for carrying out necessary diagnosis; and BDAC-5, I have connected dashboard applications or information with the manager’s communication devices.

Integration capability (IC) refers to the extent to which a firm collaborates with external upstream and downstream partners to optimise collective performance in the creation, distribution, and enhancement of product value (Flynn et al., 2010): IC-1, creating linkage with suppliers and customers through information technology; IC-2, aligning performance indicators with external partners; IC-3, real-time information sharing for making common demand forecast; IC-4, establishing strategic partnerships with external partners; and IC-5, capable to work with external partners to improve inter-organisational process.

Team collaboration (TC) focuses on the internal activities of a firm. Team collaboration was measured using a nine-item scale adopted and revised from the study of Flynn et al. (2010): TC-1, data integration among internal functions; TC-2, enterprise application integration among internal functions; TC-3, integrative inventory management; TC-4, real-time searching of the level of inventory; TC-5, real-time searching of logistics-related operating data; TC-6, the utilisation of periodic interdepartmental meetings among internal functions; TC-7, the use of cross-functional teams in process improvement; TC-8, the use of cross-functional teams in new product development; and TC-9, real-time integration and connection among all internal functions from raw material management through production, shipping, and sales.

Operational performance (OP) was measured using a seven-item scale adopted and revised from Liu et al. (2016): OP-1, decreasing product/service delivery cycle time; OP-2, rapidly responding to market demand changes; OP-3, rapidly bringing new products/services to the market; OP-4, rapidly entering new markets; OP-5, rapidly confirming customer orders; OP-6, rapidly handling customer complaints; and OP-7, establishing a strong and continuous bond with customers.


Model Fits Analysis

The maximum likelihood method and model path were used to elucidate the relationship between variables. The path coefficient was used to measure the direct influence of the latent independent variable on the latent dependent variable. The latent independent variables may have indirectly influenced the latent dependent variable through other variables. The causal structure of the hypothesised research model, which assumes linear, causal relationships among the constructs, was tested using a structural model.

LISREL analysis was performed with the theoretical model with two exogenous latent constructs, namely, IAAI, BDAC, and three latent endogenous constructs, namely, integration capability (IC), team collaboration (TC), and operational performance (OP). All the model-fitted indices of the structural model exceeded their respective common acceptance levels (Kenny, 2015; Kenny et al., 2014). The ratio of χ2 to the number of degrees of freedom was 1.171 (GFI = 0.925, NFI = 0.949, CFI = 0.72, SRMR = 0.066), and the RMSEA was 0.023 with a 90% confidence interval of 0.017–0.028, suggesting that the model fits the data well.

Test of Structural Model

Figure 7 presents the resulting structural model, path coefficients and their significance, and hypothesis test results. The SEM analysis results were as follows.

Fig. 7
figure 7

SEM structural analysis

Employees’ intention to adopt AI applications was positively related to integration capability (β = 0.23, t-value = 2.65, p < 0.01) and team collaboration (β = 0.35, t-value = 4.01, p < 0.001). The analytical results support H1a and H1b. AI has become increasingly popular and attracted wide attention of firms and has even greatly changed society and technology (Chai et al., 2021). These findings are consistent with the results of previous studies (Ajzen & Fishbein, 1980; Ajzen, 1991; Cronin et al., 2000; Jha et al., 2020).

BDAC has gained ground, thanks to its capacity to use techniques that enable managers to make better decisions based on proof rather than human judgment or intuition (Benzidia et al., 2021). BDAC was positively correlated with integration capability (β = 0.48, t-value = 4.41, p < 0.001); this finding is consistent with the results of previous studies (Rezaee & Wang, 2018; Cai et al., 2016); therefore, H2a was supported. Nevertheless, BDAC was not significantly correlated with team collaboration (β = 0.09, t-value = 1.12); thus, H2b was not supported. This result contradicts the findings of previous empirical studies (Ganbold et al., 2020; Lozada et al., 2019). The justice judgment theory and the equality rule, which suggest that all team members receive the same reward, are the dominant principle of distributive justice in a collaborative setting (Loberg et al., 2021). Nevertheless, the negative influence of BDAC on team collaboration takes effect by reducing the team members’ personal efforts and disincentivising them to share relevant information.

Integration capability was positively correlated with team collaboration (β = 0.33, t-value = 3.61, p < 0.001), supporting H3. Integration capability and team collaboration provide an environment for more in-depth understanding of project constraints and possibilities than in traditional project (Walker et al., 2017). These findings are consistent with the results of previous studies (Graham, 2018; Horn et al., 2014).

Integration capability and team collaboration were positively correlated with operational performance (β = 0.30, t-value = 3.37, p < 0.001; β = 0.41, t-value = 4.07, p < 0.001), supporting H4a and H4b. Integration capability and team collaboration provide an environment for far greater understanding of project constraints and possibilities than would occur in traditional project delivery. The analytical results supporting H4a and H4b are consistent with theoretical expectations (Walker et al., 2017; Silva et al., 2021).

Mediation Analysis

The Sobel t test (Sobel, 1982) with correction for abnormal formulation (Preacher & Hayes, 2008) was used to test the indirect effect of BDAC → integration capability → operational performance, yielding a t-value of 4.01. The t-value was statistically significant (1.96), proving that integration capability is a significant mediator in the relationship between BDAC and operational performance.

Kim et al. (2010) stressed the significance of capabilities consisting of technology, process, and people in the successful implementation of an enterprise-wide CRM strategy. Wang et al. (2018) adopted the knowledge-based theory to investigate high-tech firms in Taiwan’s science parks. They found that each dimension of knowledge networks could improve firms’ innovation performance, and firms’ knowledge integration capability has a full mediating effect on the relationship between knowledge cognition and innovation performance. Kim and Wang’s findings confirm the importance of integration capability between digitalisation capability and operational performance. The analysis results for the test mediating effects are also consistent with knowledge-based theory and their analyses result in AI applications.


Theoretical Implications

This study provided an insight into the use of AI applications in business environments and proposed both theoretical and managerial implications. First, the finding concerning the influence of integration capability and team collaboration on firms’ operational performance is consistent with previous studies (Otioma, 2022; Silva et al., 2021). Another important finding of this study is that BDAC is not positively correlated with team collaboration. Moreover, the Sobel t-value of the indirect effect of team collaboration on the relationship between BDAC and operational performance was 0.912, which is not statistically significant (1.96). This result contradicts the findings of previous empirical studies (Ganbold et al., 2020).

Acquiring BDAC is a cost-intensive process in AI applications. The following critical factor should be taking into consideration: the financial capability to purchase the latest big data analytical equipment; the willingness of employees to learn BDAC; the willingness to use BDAC; the capability, education level, and specialised field to acquire BDAC; the working environment and the interaction among co-workers; and the supplier chain. These factors together create the synergies among all stakeholders.

Team collaboration involves the integration of a large-scale supply chain that exists outside of a firm. Nevertheless, firm employees’ interactions and cooperative patterns in the supply chain are fixed, as they typically perceive their level of cooperation to be normal. Under such negative conditions, this behavioural mode may not likely to be changed over a short term. In addition, team collaboration differs from integration capability in that it involves the integrations of various customers and suppliers, whose roles may change in different applications and integration process. Furthermore, the core competencies and resources of new venture employees can affect the integration process. In the absence of proper and sufficient trainings, employees might be unqualified to utilise AI techniques and related BDA applications. The above discussion confirms that employees’ intention to adopt AI applications is positively related to integration capability and team collaboration, and BDAC is positively related to integration capability, but not to team collaboration.

The role of integration capability can change during transitions in the integration processes; external resources may become internal resources. Therefore, H3, which proposes that integration capability is positively correlated with team collaboration, is supported by the results of this study. The empirical results indicate that integration capability may alter the relationship between competitors and partners. Similar to the resources and core competencies of upstream suppliers, those of downstream sales channels and competitors may be internalised through integration. The role of integration capability can change during the transitions that occur in the integration process (Chen, 2015). In general, integration capability involves the creation of linkages with suppliers and customers through AI applications, the alignment of performance indicators with those of external partners, the sharing of information in real time for common demand forecasting, the establishment of strategic partnerships with external partners, and the collaboration with external partners to improve inter-organisational processes (Chang et al., 2016). When such integration is completed by employees, the external partners’ original requirements regarding cooperative components may become the firm’s internal resources.

Both integration capability and team collaboration are critical factors affecting operational performance. The analytical results that support H4a and H4b are consistent with our theoretical expectations (Li & Chen, 2019). With respect to the psychology of personnel, the effects of integration capability in BDAC depend mostly on the employees’ working attitude and intention, while managers cannot force employee to devise integration and collaboration through administrative policies. Furthermore, integration capability in BDAC can increase the efficiency of operational performance, while team collaboration can enhance the integration synergy, which further heightens the effectiveness of operational performance. These findings reveal the importance of integration capability and team collaboration among all team members and suppliers in achieving the expected operational performance in the AI application processes.

Managerial Implications

The results of this study indicate that employees’ intention to adopt AI applications is a principal factor in affecting the performance of integration capability and team collaboration. The success or failure of employees’ intention to adopt AI applications not only relies on administrative decrees, formal orders, or procedures. On the other hand, the formation of behavioural intentions includes cognitive factors, emotional factors, behavioural factors, and loyalty. The establishment of well-planned training to enable employees for realising the advantage to use AI applications affects the success of prompting integration capability and team collaboration. In practice, the firm should consider job applicants exhibiting stronger intention to adopt AI applications, solid background in AI training/education, and/or familiarity with AI applications.

AI is often regarded as the field of computing, but indeed involves two aspects, namely, training (machine learning) and inferencing. For any AI system to work, a neural network must be first trained. Training requires intensive computing operations, such as feed-forward and back-propagation, where logic cores are fed copious amounts of data (Cheng, 2019). Hence, digitalisation is crucial and closely related to AI applications.

This study yielded important practical implications for managers who seek to optimise the positive relation between BDAC and team collaboration as well as design effective incentive systems. Perceived injustice may have negative effects that are not discussed herein, such as increased turnover intentions. Given the increasing competition for skilled employees, perceived injustice can be particularly damaging to a firm’s ability to attract and retain productive employees and, in turn, to its long-term success. Although BDAC was not positively correlated with team collaboration, as indicated in this study, integration capability and team collaboration were significantly correlated with operational performance, suggesting that integration capability and team collaboration together have a synergistic effect on operational performance. Ceccagnoli et al. (2012) examined whether participation in an ecosystem partnership could improve the business performance of small independent software vendors in the enterprise software industry and how appropriability mechanisms influence the benefits of partnership. Their finding emphasises the importance for capability of integration and team collaboration, especially in the ERP applications of cross-departmental integration. Gardner et al. (2012) studied the direct and integrated effects of sets of capability strengths on competitive advantage, and its empirical correlation with relative performance. Furthermore, they explored how teams can develop a knowledge-integration capability to dynamically integrate members’ resources into higher performance. Furthermore, Woo et al. (2016) indicated that suppliers with higher information sharing capabilities have better team collaboration, contribute to cost reduction, and achieve competitive advantages. Their empirical results demonstrate the importance of integration capability and team collaboration in operational performance. In the management of AI applications, BDAC can be implemented in parallel with employees’ strong intention to adopt AI applications, which is more effective than implementing them sequentially. Employees’ intention to adopt AI applications and the strategic implementation of BDAC can enable a firm to improve its operational performance when driven by the integration capability and team collaboration to use AI, which further enhances operational performance. These processes indicate that employees’ intention to adopt AI applications and BDAC are both crucial for the success of a firm’s integration capability and team collaboration and for effective management to improve operational performance in the face of global competition.

Under the trend of business ecosystem, the integration capability is becoming crucial for business success. The statistical results of this study suggest that BDAC is positively correlated with integration capability, thus supporting H2a, as well as H4a, and the mediating effect between BDAC and operational performance. Firms that employ well-trained employees with BDAC can effectively analyse the marketing trend and consumer demands from diversifying perspectives efficiently, thereby significantly improving the organisational integration capability from different resources. This in turn provides great contributions to management decisions, which further enhance the operational performance. As for high-tech firms, this study suggests that AI technology, which relies on employees’ intention to adopt AI applications, BDAC, integration capability, and team collaboration, should be promoted in firms to enhance their operational performance. In addition, AI technology can help firms to realise their business goals through integration capability and team collaboration, thereby enhancing operational performance.

Ideas for Further Research

This study filled in the research gap on employees’ intention to adopt AI application and BDCA that influence the integration capability and team collaboration and the effect on firms’ operational performance. The results of this paper confirm that in workplaces, employees’ intention to adopt AI applications, and BDAC can improve firms’ operational performance and contribute to fulfilling organisational goals. In addition, AI technology can improve operational performance by facilitating the realisation of goals through integration capability and team collaboration. Thus, AI-based applications have a critical influence on operational performance.

Although this research makes several crucial contributions, some directions for future research are proposed. First, regarding the use of SEM, most hypotheses were confirmed, and both the constructs and the model were validated. Because this was a first attempt to organise numerous factors simultaneously while exploring their correlations, this study was exploratory to some extent. Therefore, future studies can develop new scales for these constructs to provide more elaborate measurements. Second, further study on different industries/products is necessary because different products have dissimilar product life cycles and rely on different team collaboration patterns. Finally, time structures should be considered. A panel analysis with measurements at different time points may reveal relationships among variables. Furthermore, the analysis on other organisational performance, for example, logistic management performance, new product performance, or financial performance, can be further studied.