This study investigates the role of generative artificial intelligence (GAI) in enabling sustainable organizational transformation. Based on an empirical analysis of survey data collected from 424 employees across companies in Heilongjiang, Shandong, and Guangdong Provinces in China, this study examines how knowledge sharing (KS) and employee attitudes toward AI may moderate the adoption of GAI and how digital leadership (DL) reinforces these relationships. The results indicated that GAI enabled internal KS and allowed employees to adopt more positive attitudes toward digital or AI tools, thereby contributing to better environmental performance that ensured long-term organizational sustainability. DL can moderate across technology, understanding, and performance. The leaders who interpret the trend in technology, understand the effect of that technology on organizational processes, and can link those processes to performance goals in order to create circumstances propitious for introducing GAI. Within this context, workers would be well-positioned to understand the potential of digital technology and collaborate in KS as well as behavioral change toward achieving goals of sustainability. With the Technology‒Organization‒Environment theory and the Resource-Based View theory in focus, a two-path model linking GAI adoption and sustainable performance is proposed in this study. The results offer valuable suggestions for organizations concerning the adoption of digital technology in environmental objectives for improved long-term sustainable performance.
Recently, the domain of operational control is undergoing a radical shift because of the emergence of generative AI (GAI). Nowadays, many firms are including GAI implementation within their administration and operations for enhanced knowledge processing abilities, motivating employees’ creativity, and overcoming various business challenges through information-based insight (Li et al., 2025). The user base of ChatGPT reached 100 million daily active users. It broke a record with the speediest growth among all consumer applications so far. This recognizes GAI’s huge potential for influence (Dwivedi et al., 2023). GAI can change the nature of knowledge work and business models. It will provide unmatched potential for productivity and innovation and will also generate a new set of intensive needs (Thomas et al., 2024). As indicated by higher-level research, GAI implementation will hopefully allow organizations to derive a competitive advantage, build a paradigm shift in green innovation, and ensure sustainable development outcomes (Kassa & Worku, 2025). In many organizations, these advantages increasingly depend on the effective use of digital technologies to support knowledge-intensive and service-oriented organizational activities, instead of solely on traditional operational efficiency (Paiola et al., 2024; Rapaccini et al., 2023). Despite this, organizations have still not been able to unlock its potential while minimizing potential threats (Badghish & Soomro, 2024). The effects and effectiveness of GAI on control within an organization need to be analyzed.
Though there have been significant discussions about the possible effects of GAIs on an organization’s effectiveness, previous research has been conducted mostly on its possible application based on a micro-level approach. Nevertheless, no conclusive findings exist on the internal processes that may be used by GAIs to ensure organizational performance over time (Florea & Croitoru, 2025). Knowledge sharing (KS) is considered to be one of the essential processes that may boost innovation and performance; nonetheless, it is unclear yet regarding the procedure followed by the application of GAI within employee KS processes (Qahl & Yahya, 2024). In organizations where value creation increasingly relies on collaborative knowledge use and internal service processes, understanding how GAI influences KS is particularly important (Chirumalla et al., 2023). Similarly, user attitude toward AI (AT AI; e.g., acceptance or trust in AI) is sometimes driven by the perceptions of benefits associated with technology ethics (Rasheed et al., 2024) and can be a determining factor for successful or unsuccessful implementation of GAIs, but the mediating role of psychology for fruitful performance by GAIs in organizations has not yet been proven (Neiroukh et al., 2025). Moreover, digital leadership (DL) is a prominent contextual factor for organizations adapting to technology (Verhoef et al., 2021). DL may shape how employees interpret and apply GAI in knowledge- and service-related work activities, yet the internal mechanisms and boundary conditions through which GAI influences sustainable organizational performance (SOP) remain unclear and require further investigation.
The adoption of new or emerging technologies could be considered a strategic resource. The integrated resource-based view (RBV) theory provides a framework on how such resources, if invested with the right skills in the organization, could provide competitive advantages and subsequent long-term performance (Hui et al., 2025). Dynamic capability approach focuses on an organization’s innovation capabilities and its adaptive response to changes in its environment, requiring us to think about how the organization dynamically reshapes its knowledge and capabilities when it adopts GAI technology (Singh et al., 2024). A such, KS becomes a vital link for technology adoption and value development from a knowledge management lens (Gazi et al., 2024; Gong et al., 2025). From this perspective, GAI can support organizational value creation by enabling employees to recombine knowledge and improve internal service efficiency, which is essential for sustainable performance (Luo & Liu, 2024). The green knowledge management practices within the organization might also influence the link between capabilities in AI technology and sustainability for the environment (Kumar et al., 2025). Hence, the adoption of the RBV theory—the concept of dynamic capabilities—together with the practices in knowledge management will help in comprehending GAI capabilities in enhancing the organization’s sustainability performance in economic, ecological, and societal dimensions using KS practices and user attitudes in knowledge management (Lin, 2025). Furthermore, using DL as a moderating variable helps to understand the relationship among "technology, understanding, and performance" and explore the influence of organizational frameworks on the aforementioned mechanisms.
This study offers a strategic approach for organizations embracing GAI and examines its organizational implications. The findings reveal the significance of knowledge integration and attitude toward realizing innovation capabilities from the perspective of AI. Recommendations include training programs, usefulness, and ethics. DL is identified as an important enabler for the development of digital skills to offset risks related to AI. By focusing on internal knowledge processes and employee attitudes, this study highlights how GAI adoption (GAI A) can enhance SOP in knowledge- and service-intensive work contexts. A multi-level model was tested. GAI A directly improves KS and attitudes, which help in achieving sustainable organizational outcomes. KS and attitudes act as the two mediating variables, and DL is the moderating variable. This study adds to the current research on the adoption of AI in the modern world of digital culture.
Theoretical background and hypothesis developmentGAI A and KSGAI A is the phenomenon of people, organizations, or enterprises integrating and applying the GAI technology to their work practices and daily operations, product and service development and innovation, and knowledge and skills development and training (Venkatesh et al., 2003). It usually involves technical feasibility studies, system integration, user training programs and formulating policies and behavioral guidelines to improve efficiency, self-reliant content creation, and innovation support, an important trend of AI technology development and practical applications (Ooi et al., 2025). KS is an individual or business act of transferring knowledge, experience, and information from within and/or from one organization to outsiders with various forms and purposes, playing an important role in enhancing innovation capabilities and improving work performance (Wang & Noe, 2010). Organizational environment, technical support, and incentive systems are viewed as crucial elements that influence KS intentions and behavior (Bartol & Srivastava, 2002; Kim & Lee, 2006). KS becomes an important topic related to corporate strategies and personnel management in the knowledge-based economy.
There has been increasing scientific interest in recent years in embracing GAI A in operational settings, particularly its potential in fostering knowledge dissemination and innovation. By promoting the timely dissemination of information, it has been proved that GAI A enhances real-time communication and makes teamwork more efficient (Ola-Oluwa, 2024). The integration of knowledge dissemination in an organizational setup is facilitated by this technological advantage (Khan et al., 2025; Zamrudi et al., 2025). Findings by Qahl and Yahya (2024) reveal that a need for knowledge dissemination for implementing GAI A in higher learning institutions. Hence, we proposed the following hypothesis.
H1 GAI A has a positive impact on KS.
AT AI is described as the overall appraisal, emotional experience, and behavior tendency of people or organizations toward the technology of AI, which can be either negative, such as fear and discomfort, or positive, such as trust and anticipation (Long & Magerko, 2020). This attitude is anticipated to play an essential role in influencing the public acceptance and willingness to use the technology of AI (Shin, 2021).
There has been a growing attitude change among people regarding AI systems driven by the increased adoption of GAI A technologies. The interactivity, personalization, and ease of use of GAI A (Gupta et al., 2024) contribute primarily to this positive feedback. The widespread use of GAI A has effectively reduced the technical anxiety of the teacher group and increased their acceptance of AI technology (Nikolic et al., 2024). When GAI users perceive trust in the usefulness and ease of use of the technology, their willingness to continue using it will ultimately increase (Vorobeva et al., 2024). Thus, we can formulate the following assumption.
H2 GAI A has a significant direct positive impact on users’ AT AI.
SOP means the ability of an organization to achieve long- term, balanced, and stable growth regarding economic benefits, environmental responsibility, and social impact (Elkington & Rowlands, 1999). It emphasizes brief financial results and connects importance to resource efficiency, individual well- being, and social benefit (Epstein, 2018). According to Lozano (2015), successful proper planning and responsible practices are essential for enhancing an organization's sustainability efficiency. This report adopts a 5-point product range developed by Kordab et al. (2020).
KS has grown to be a significant driving force for enhancing organizations' long-term efficiency in a time when both digitalization and sustainable development are crucial. KS promotes organizational creativity and green innovation by strengthening the flow of information within the organization, thereby significantly improving the latter's sustainable performance (Gazi et al., 2024). By promoting the inheritance of experience and creative exchange among organizational members, KS effectively promotes organizations' innovation capabilities and green practices, ultimately improving their environmental and organizational performance significantly (Malik et al., 2024). KS not only improves the individual capabilities of employees but also promotes the realization of SOP through synergy at the organizational level (Olan et al., 2024). Hence, we proposed the following hypothesis:
H3 KS has a positive impact on SOP.
As AI systems become increasingly integrated into operational options, employees' AT AI have been recognized as a crucial parameter controlling its commitment to organizational performance. Based on Hu et al. (2023), there is evidence that people who have good AT AI have a great level of interest in adopting the system, which assists in improving organizational performance in terms of environmental and operational conditions. According to Womick (2024), attitude has been a critical element to consider in terms of people's desire to adopt AI applications. In general, people's AT AI are supported by their effort and innovation spirit, which have served in favor of achieving objectives for organizational sustainability. Moreover, people's optimistic attitude toward AI adoption has been supportive in building their working relations, thus contributing toward overall improvement in organizational performance. This fact has been evidenced in Khan et al. (2025). Accordingly, the following hypothesis is developed:
H4 Users’ AT AI have a positive impact on SOP.
The advent of GAI adoptability has created a massive transformation chance for organizations, and it has also witnessed considerable potential in extending the limits toward sustainable performance (Badghish & Soomro, 2024). Along with enhancing the abilities of the organizations in a manner that it innovates in exploitative and exploitive capacities, GAI adoptability also facilitates improved performance of the organizations at a later stage (Singh et al., 2024). Along with assisting in task execution, creative development, and automating processes, adopting collaborative GAI (Przegalinska et al., 2025) also assists with enhancing the working efficiency and performance of organizations (Przegalinska et al., 2025). Therefore, the choice to adopt GAI in organizations is both professional and an important step toward sustainable development. Thus, the following hypothesis is proposed.
H5 GAI A has a positive impact on SOP.
With the increased adoption of GAI in different institutions, the actual benefits can only be attained with the assistance of a KS system (Kumar et al., 2025). It is evident that with a KS culture in institutions, the potential for innovation in the green area introduced by the competencies created by AI can be tapped to enhance sustainability (Gazi et al., 2024). KS in an AI-based organizational culture—another important element that bridges the gap between two types of knowledge with the capacity to improve performance. That is, KS is the necessary linkage between the application of technology and the achievements in green performance. The complex base and the individual behavior are crucial for the successful implementation of GAI A by organizations. Employers who are more open to and use GAI A are more likely to do so in large numbers, which increases organizational innovation efficiency and sustainability (Ghimire et al., 2024). Additionally, if employees have a positive AT AI, it will be simpler to overcome the fear and uncertainty surrounding the adoption and, in the end, enable the GAI A's potential to improve (Goochee & Dumont, 2024). Hence, the following hypotheses are developed:
H6a KS positively mediates the relationship between GAI A and SOP.
H6b Users’ AT AI positively mediate the relationship between GAI A and SOP.
DL means the ability of leaders to use advanced technologies to promote organizational change, innovation, and performance improvement (El Sawy et al., 2020). To help people react to the rapidly evolving scientific setting (Kane, 2019), this leadership requires modern thinking, modern sensitivity, and change management skills. In order to realize corporate transformation plans and maintainable competitive advantages, DL is essential (Zeike et al., 2019).
DL is an important component of creating a KS lifestyle when putting GAI into practice in organizations. Effective GenAI implementation depends on officials with a forward- seeking vision and strong technological ability who can encourage people to embrace new resources and collaborative techniques (Shields, 2024). Through a systematic literature review, Paiuc and Iliescu (2022) emphasize that DL can reshape organizational knowledge management culture, enhance employees' cultural intelligence, and improve the efficiency of knowledge circulation. Under conditions of strong DL, the introduction of GenAI is more likely to stimulate knowledge exchange among organizational members, thereby promoting collaborative learning and innovation (Mohammad et al., 2025). Digital leaders can effectively increase employees' acceptance of new technologies and encourage the widespread adoption of GAI A (Radoui & Cherradi, 2025). During GenAI implementation, the DL demonstrated by organizational leaders helps reduce employees ' uncertainty and resistance to technology, improving both their cognitive acceptance and emotional response to AI (Zhang et al., 2024). By introducing knowledge dynamics and cultural intelligence, DL not only guides the organization's strategic direction but also cultivates a positive learning climate, thereby strengthening employees' identification with new technologies (Dwivedi et al., 2023). Hence, the following hypotheses are developed.
H7a DL positively moderates the relationship between GAI A and KS.
H7b DL positively moderates the relationship between GAI A and users’ AT AI.
The Technology‒Organization‒Environment model discusses how internal organizational characteristics and situational factors shape the implementation process for GAI (DePietro et al., 1990). In addition to the RBV theory (Barney, 1991), sharing knowledge and the capabilities of AI can be regarded as strategic organizational resources that help in sustaining organizational performance. Therefore, grounded by the above theoretical concepts and the purpose of this study, the conceptual framework for understanding the process by which SOP outcomes are shaped by GAI implementation GAI (Fig. 1) has been proposed.
Research methodologySample and data collection
This study used a structured survey methodology via a questionnaire to explore the inter-connections of adopting GAI, digital transformation of leaders, KS, and SOP. Using a survey methodology is effective for researching employees' views and attitudes toward business practices and technology usage, including emerging digital technologies. This survey took place anonymously among adults, offering minimal risk to the participants. Introduction to the survey took place on the first page of the survey instrument, and it included information about study objectives, its purely voluntary nature, ensuring the confidence and anonymity of all data provided, and participants' rights. Consent to study participants was established via an "opt-in" method; only those giving permission to participate were allowed to carry out the survey process, which refused all participants who did not want to participate, preventing any record or data to be registered. The participants also had the right to choose to stop at any point during the survey without any consequence. No personal identifying information was collected, and all the data were analyzed in anonymous form. Data were collected for full-time company employees working in three regions of varying economic structures and industries in southern, northeastern, and central China: Guangdong, Heilongjiang, and Shandong Provinces, respectively. These regions include a varied economic and industry mix, which can help to eliminate any region-related bias. As there was no total sampling frame of employees who use GAI tools, convenience sampling was used for data collection. This method has limitations, but it is widely accepted and used for studies about organizations and information systems, including focusing on emerging technology.
A total of 450 online questionnaires were distributed, and 435 were received. To add some relevance to the data as well as to improve the levels of validity that count, a screening question was added to the initial stages of the research. This targeted the respondents in terms of their experience in utilizing the power of GAI. Those who answered that they did not possess such experience were removed when considering the final analysis. This ensured that only individuals who possessed experience were left, hence raising the substantive levels of validity for the measures implemented within the study. A final count of 424 valid questionnaires was obtained, thus ensuring that the response rate was 94.2 %. Table 1 highlights the demographic factors considered, which showed the respondents being representative of diverse demographic features with regard to gender, age, educational attainments, work experience, industry, level, and company size. More precisely, 51.2 % of the respondents identified themselves as male, and 48.8 % identified as female. The largest share of respondents belonged to younger age groups: 47.4 % aged 21‒30 and 29.7 % aged 31‒40. In terms of education level, >70 % of the participants had reached at least a bachelor’s degree; in all, 53.5 % had a bachelor’s degree, 15.8 % a master’s degree, and 2.1 % a doctoral degree. The largest industry sectors included >1‒10 years of work experience: 61.2 %. In industry terms, most respondents worked within the trade industry: 28.8 %. More precisely, 80.9 % of the respondents were administrative/office staff, 14.9 % first-line managers, but only 4.2 % mid-senior managers. In terms of company size, small to medium-sized enterprises represented most of the observed sample: 46.0 % of the respondents worked for organizations employing 51‒200 people, whereas 26.2 % worked for 201‒500 people-sized firms. Particularly, 39.9 % of the organizations surveyed have been investing "high/very high" into AI research, indicating a rather enthusiastic attitude toward digital transformation. Summarily, the observations show that the observed sample comprises rather young, highly educated, and practically experienced staff working predominantly in administrative/office staff roles.
Demographic information.
Notes. N = 424
In this study, the relationship between GAI A, KS, AT AI, SOP, and DL is explored using appropriate measurements. A scale for GAI A is designed with a set of four scaled questions by Zamrudi et al. (2025). Sample items include "I often use generative AI tools at work," and "My organization encourages employees to explore and use generative AI tools." KS is measured by a four-item scale created by Jacobs and Roodt (2007), which gauges the willingness and frequency of KS activities among co-workers. AT AI are measured through a five-item scale designed by Sindermann et al. (2021). It measures individuals’ desire and attitude to learn from and work with AI in their respective working settings. SOP is measured through a five-item scale designed by adapting, to some extent, the ones designed by Kordab et al. (2020). The theme here is centered on the participation of organizations within the modern market and relative efficiency in running. Finally, DL is measured through a six-item scale designed by Zeike et al. (2019). This measures digital vision, leaders’ support for digitalization initiatives, and their competencies in utilizing emerging technologies such as AI and Cloud Computing in their organizations. The measures of these variables are done through a 5-point Likert scale ranging from 5 (Strongly Disagree) to 1 (Strongly Agree).
Internal consistency, indicator loadings, and construct reliability were examined to determine the reliability and validity of the measurement model. The results in Table 2 reveal that all constructs have been found to possess good levels of internal consistency. The value of Cronbach's alpha is ranging from 0.827 to 0.916, which is above the recommended threshold of 0.70 (Hair et al., 2010). Standardized factor loadings verified convergent validity, and all loadings exceeded the minimum acceptable value of 0.70. Specifically, the item loadings of each latent construct were: KS 0.744 to 0.816; AT AI 0.767 to 0.797; GAI A 0.774 to 0.810; SOP 0.745 to 0.795; DL 0.744 to 0.813. These findings show that each item has strong reliability for its respective construct. To sum up, the above findings show that the model has high internal consistency reliability and high convergent validity, providing a strong basis for testing of the structural model.
Exploratory factor analysis results.
Notes. N = 424.
As all the key variables were measured on the same set of respondents by using the same survey instrument, CMB could be a potential problem for consideration. To deal with the problem, Harman’s single-factor test was performed with exploratory factor analysis. Looking at Table 3, the first unrotated factor explained only 37.39 % of the variance, which was less than the acceptable limit of 40 %. Thus, the CMB problem is not a significant issue for consideration in the current research.
Common method bias and diagnostic tests.
Notes: VIF values were calculated based on a multi-variable regression model including GAI adoption, KS, AT AI, and DL.
Further, confirmatory factor analysis (CFA) was performed to compare and contrast the proposed multi-factor model of measurements with a single-factor model. The findings showed significantly greater fit of the multi-factor model, which again helped to mitigate any potential concerns with common method variance.
We also examined multicollinearity by calculating variance inflation factors (VIFs). All VIF values were below commonly accepted thresholds, indicating that multicollinearity was not a concern. Furthermore, skewness and kurtosis values fell within acceptable ranges, supporting the assumption of approximate normality required for the structural equation model (SEM) estimation.
Results and analysisDescriptive statistics and confirmatory factor analysisTo test the convergent validity and construct validity, this study used AMOS 26.0 to conduct CFA. Table 4 shows the results of CFA. The model fit index shows that the overall measurement model has a good fit (CMIN/DF = 1.149, GFI = 0.949, CFI = 0.994, TLI = 0.993, RMSEA = 0.019, RMR = 0.038), which meets the model fit criteria recommended by scholars such as Hu and Bentler (1999). From the results, the loadings of all factors are very significant (p < 0.001), providing strong evidence for the hypothesized construct relationship. The combined reliability scores of GAI A (0.824), AT AI (0.854), KS (0.830), SOP (0.851), and DL (0.879) all exceed the commonly used threshold of 0.70, indicating strong internal consistency. In addition, the average variance extracted (AVE) of these constructs was also above the recommended level of 0.50 (GAI A = 0.609; AT AI=0.614; KS=0.616; SOP=0.599; DL=0.638), indicating adequate convergent validity. Fornell and Larcker's (1981) criteria are used to assess discriminant validity, which compares each construct's AVE's square root to its shared variance. This analysis ensured that the variance shared by each construct in the model with its indicators was greater than that shared with other constructs, thus further verifying the model's validity.
Confirmatory factor analysis results.
Model fit: CMIN/DF = 1.149, p < 0.001, RMR=0.038, GFI=. 949, CFI= 0.994, TLI= 0.993, IFI = 0.994, RFI=0.947, NFI=0.953, RMSEA= 0.019.
Notes. N = 424, Gen AI A=Generative AI Adoption, KS=Knowledge Sharing, AT AI=Attitude Toward AI, SOP=Sustainable Organizational Performance, DL=Digital Leadership.
Prior to conducting structural equation modeling, this study conducted descriptive statistics and correlation analysis. Table 5 presents each variable's mean, standard deviation (SD), and Pearson correlation coefficient, which is used to evaluate the basic relationship between variables and preliminarily test potential multicollinearity problems. From the descriptive statistical results, the mean of adoption of GAI A is 3.328 (SD = 0.972), and that of AT AI is 3.338 (SD = 0.971), reflecting that the respondents have an overall positive attitude toward the use and cognition of AI-related tools. The values of means pertaining to KS, SOP, and DL are 3.406, 3.324, and 3.300, respectively, while those of the SD range from 0.942 to 1.008. This reveals that the values of the variables tend to be reasonable. The result of the correlation analysis reveals that the values among most variables tend to be significantly positive. Specifically, there is a significant positive correlation between GAI A and AT AI (r = 0.440, p < 0.001). GAI A also has a significant positive correlation with KS (r = 0.378, p < 0.001), SOP (r = 0.361, p < 0.001), and DL (r = 0.420, p < 0.001), suggesting that GAI A is significant in relation to variables such as AT AI, KS, SOP, and DL. Additionally, KS is greatly and positively correlated with SOP (r = 0.425, p < 0.001) and DL (r = 0.457, p < 0.001), which again supports the theoretical hypothesis that it is a mediation variable. The positive correlation of AT AI with SOP (r = 0.436, p < 0.001) and DL (r = 0.383, p < 0.001) is also significant. Concerning demographic variables, the high correlation among age and work experience (r = 0.830, p < 0.001) supports logical reasoning. However, the correlation among gender-related variables is insignificant, indicating that gender has little bearing on the significant variables identified. All the correlations among the identified variables failed to surpass 0.85, which preliminarily indicated the absence of severe multicollinearity concerns among the identified variables with a solid foundation for analysis via a structural model.
Descriptive statistics and correlation matrix.
| Variables | Mean | S.D. | Gen AI A | ATAI | KS | SOP | DL |
|---|---|---|---|---|---|---|---|
| Gen AI A | 3.328 | 0.972 | 0.780 | ||||
| ATAI | 3.338 | 0.971 | .440⁎⁎ | 0.784 | |||
| KS | 3.406 | 0.970 | .378⁎⁎ | .405⁎⁎ | 0.785 | ||
| SOP | 3.324 | 0.942 | .361⁎⁎ | .436⁎⁎ | .425⁎⁎ | 0.774 | |
| DL | 3.300 | 1.008 | .420⁎⁎ | .383⁎⁎ | .457⁎⁎ | .441⁎⁎ | 0.799 |
Notes. N = 424 *p< .05.
To verify hypotheses H1 to H5, a SEM was built with the assistance of AMOS 26.0 for path analysis (Fig. 2). The results are presented in Table 6. The SEM technique is quite efficient in determining the association between GAI A, KS activities, AT AI systems, and SOP metrics. Standardized path values, critical ratios (CR), along with the significance values (p-value), explain the significance of the modeled association between the variables. For the entire structural model, the fitness values explain its relevance along with the significance of the modeled paths with good support for the values: CMIN/DF = 1.382, p-value < 0.001, RMR = 0.072, GFI = 0.956, CFI = 0.988, TLI = 0.985, IFI = 0.988, NFI = 0.956, RMSEA = 0.030. These values satisfy the criteria suggested by Hu and Bentler (1999) for establishing a proper fitness for the entire structural model.
Structural equation modeling results for hypotheses testing.
| Hypothesized path | Estimate | S.E. | C.R. | P | Results |
|---|---|---|---|---|---|
| H1. Gen AI A ➝ KS | 0.437 | 0.055 | 7.881 | ⁎⁎* | Supported |
| H2. Gen AI A ➝ ATAI | 0.542 | 0.059 | 9.182 | ⁎⁎* | Supported |
| H3. KS ➝ SOP | 0.310 | 0.060 | 5.122 | ⁎⁎* | Supported |
| H4. ATAI ➝ SOP | 0.279 | 0.057 | 4.882 | ⁎⁎* | Supported |
| H5. Gen AI A ➝ SOP | 0.140 | 0.066 | 2.134 | ⁎⁎ | Supported |
Model fit:.
CMIN/DF = 1.382, p < 0.001, RMR=0.077, GFI=. 956, CFI= 0.988, TLI= 0.985, IFI = 0.988, RFI=0.949, NFI=0.956, RMSEA= 0.030.
Notes. N = 424.
In particular, Hypothesis 1 suggests that GAI implementation exerts a positive impact on KS, with a path coefficient of 0.437 (SE = 0.055, CR = 7.881, p < 0.001). This finding illustrates that introducing GAI in an organization exerts a positive effect on KS. Hypothesis 2 believes that GAI A will positively affect AT AI, with a path estimate of 0.542 (SE = 0.059, CR = 9.182, p < 0.001), verifying the positive guiding role of GAI use behavior on employee technology attitudes. Hypotheses 3 and 4, respectively, verify the positive impact of KS (β = 0.310, p < 0.001) and AT AI (β = 0.279, p < 0.001) on SOP, supporting the positive role of KS and attitude variables in SOP. Finally, Hypothesis 5 proposes that GAI A directly affects SOP, with a path coefficient of 0.140 (SE = 0.066, CR = 2.134, p = 0.033). The result reached a statistically significant level, indicating that GAI A affects SOP through a mediating path and has a direct mechanism.
The mediating effects of KS and AT AI in the relationship between GAI A and SOP were examined using structural equation modeling with bias-corrected bootstrap procedures (5000 resamples, 95 % Confidence Interval (CI)). To additionally reinforce the strength of the mediation analysis, the phantom variable method was used in the SEM in order to provide the estimates of the indirect effects. As the inclusion of the phantom variables provides the facility to model the indirect effects whereas the original paths remain unaffected, it is very effective in association with the bootstrap test in order to make appropriate inferences concerning the mediation effects without having any impact on the original model (Fig. 3). The model showed an acceptable fit (CMIN/DF = 1.382, RMR = 0.077, GFI = 0.956, CFI = 0.988, TLI = 0.985, RMSEA = 0.030). According to Table 7, the indirect effect of GAI A on SOP via KS (H6a) was statistically significant. The estimated value of the indirect effect was 0.132 (SE = 0.030), with a 95 % bias-corrected CI of (0.078, 0.198) not containing zero. Next, the indirect effect of GAI A on SOP via AT AI (H6b) was significant. The estimated value of the indirect effect was 0.147 (SE = 0.036), with a 95 % bias-corrected CI of (0.083, 0.225). Taken together, these results indicate that both KS and AT AI function as significant mediating mechanisms linking GAI A to SOP. This suggests that knowledge-related and attitudinal mechanisms represent important, though not exclusive, pathways through which GAI A contributes to SOP.
Regression analysis of mediating effects.
| Indirect Effects | hypothesized path | Estimate | S.E. | P | 95 % CI | Results | |
|---|---|---|---|---|---|---|---|
| Lower | Upper | ||||||
| H6a | Gen AI A→KS→SOP | .132 | 0.030 | ⁎⁎⁎ | .078 | .198 | Supported |
| H6b | Gen AI A→ATAI→SOP | .147 | 0.036 | ⁎⁎⁎ | .083 | 225 | Supported |
| Model Summary | CMIN/DF = 1.382, p < 0.001, RMR=0.077, GFI. 956, CFI= 0.988, TLI= 0.985, IFI = 0.988, RFI=0.949, NFI=0.956, RMSEA= 0.030 | ||||||
Notes. N = 424.
Table 8 presents the moderation analyses results of exploring whether DL moderates the relationship between GAI An and two mediators: KS (H7a) and AT AI (H7b).
Moderating effects of digital leadership.
| H7a | predictor variable | b | t | p | LLCI | ULCI | R-sq | F |
| ⁎⁎⁎ | 0.265 | 50.604⁎⁎⁎ | ||||||
| Gen AI A | 0.202 | 4.321 | ⁎⁎⁎ | 0.110 | 0.293 | |||
| DL | 0.331 | 7.390 | ⁎⁎⁎ | 0.243 | 0.419 | |||
| Gen AI A * DL | 0.119 | 2.892 | ⁎⁎ | 0.038 | 0.200 | 0.015 | 8.362⁎⁎⁎ | |
| Lower DL | 0.081 | 1.204 | 0.229 | −0.052 | 0.214 | |||
| High DL | 0.322 | 5.660 | ⁎⁎⁎ | 0.210 | 0.434 | |||
| H7b | ⁎⁎⁎ | 0.289 | 56.919⁎⁎⁎ | |||||
| Gen AI A | 0.297 | 6.451 | ⁎⁎⁎ | 0.206 | 0.387 | |||
| DL | 0.199 | 4.515 | ⁎⁎⁎ | 0.112 | 0.286 | |||
| Gen AI A * DL | 0.216 | 5.315 | ⁎⁎⁎ | 0.136 | 0.296 | 0.048 | 28.250⁎⁎⁎ | |
| Lower DL | 0.079 | 1.184 | 0.237 | −0.052 | 0.210 | |||
| High DL | 0.514 | 9.180 | ⁎⁎⁎ | 0.404 | 0.625 |
Notes. N = 424.
In support of H7a, the interaction between GAI A and DL is observed to be statistically significant (β = 0.119, p = 0.004), indicating that DL strengthen the positive impact of GAI adoption on KS. Simple slope analysis further revealed that the relationship between GAI A and KS was stronger at high levels of DL (β = 0.322, p < 0.001) than at low levels (β = 0.081, p = 0.229), suggesting a significant conditional effect only under high DL. The model explained 26.5 % of the variance (R² = 0.265, F = 50.604), with the interaction term contributing an additional 1.5 % (ΔR² = 0.015, F = 8.362). Similarly, for H7b, the interaction term GAI A × DL is also significant (β = 0.216, p < 0.001), confirming that DL positively moderates the effect of GAI A on AT AI. At high levels of DL, the relationship was considerably stronger (β = 0.514, p < 0.001), while it was non-significant under low DL (β = 0.079, p = 0.237). The model accounted for 28.9 % of the variance (R² = 0.289, F = 56.919), with the interaction term increasing the explained variance by 4.8 % (ΔR² = 0.048, F = 28.250).
To further verify Hypotheses 7a and 7b, this study visualizes the moderating effect based on the R language. It draws the three-dimensional regression surface diagrams shown in Figs. 3 and 4, respectively, to show the changing trend of the impact of GAI A on KS and AT AI under different levels of DL. In the figure, the horizontal and vertical axes are the levels of GAI A and DL after centralization, respectively, and the vertical axis corresponds to the predicted scores of KS (Fig. 4) and AT AI (Fig. 5). The image constructs a prediction grid using a linear regression model containing interaction terms. It is generated by the Plotly package, which intuitively shows the form of the moderating mechanism. The results in Fig. 3 show that DL positively moderates the impact of GAI A on KS. When the DL is high, the promotion effect of GAI A on KS is more significant. When the DL is low, the influence trend tends to be flat or even weakened, verifying the validity of Hypothesis 7a. The results in Fig. 4 further show that DL also positively moderates the impact of GAI A on AT AI. The higher the DL level, the more GAI A improves employees' AT AI; when the DL level is low, this relationship weakens or even partially shows a negative trend. This result supports Hypothesis 7b and underlines the pivotal role of digital leaders in emerging technology adoption.
Conclusions and implicationsTheoretical contributions
Theoretically, this study integrates several theoretical approaches into a new model to further explore the mechanism by which GAI promotes SOP. First, based on RBV, this research considers the GAI A as a strategic resource that could enable companies to establish long-term advantages in competition and improve SOP; it also proves that this can create competitive advantages and enhance SOP under the condition of matching appropriate organizational capabilities (Hui et al., 2025). Embedding the power of AI tools in core business processes helps an organization to uplift its efficiency, innovation capabilities, and adaptability to market uncertainties. Importantly, this study extends RBV by highlighting how GAI contributes to value creation through internal organizational processes instead of solely through technological investment, thereby reinforcing its relevance in service-oriented and knowledge-intensive organizational contexts (Shen & Badulescu, 2025).
Based on the dynamic capability theory, this study emphasizes the role of organizations continuously reconstructing knowledge and capabilities to environmental changes in a changing environment (Singh et al., 2024). In this context, GAI is an important enabler of organizational agility, with the application of AI advancing dynamic innovation capability. From this perspective, GAI supports organizations in transforming dispersed knowledge into repeatable and scalable organizational practices, which is particularly critical for sustaining long-term performance in digitally enabled and service-driven operational settings (Li et al., 2025; Mariani & Dwivedi et al., 2024). In addition, based on the knowledge management aspect, the research has verified the effect of KS in connecting technology application and value creation as a significant bridge (Gazi et al., 2024; Gong et al., 2025).
This study integrates RBV, the dynamic capability theory, and knowledge management perspectives to develop a conceptual framework covering factors such as GAI A, KS behavior, AT AI, and DL that explains how emerging technologies enable corporate overall improvement of performance through knowledge paths and psychological paths (Lin, 2025). This integrated framework contributes to the literature by clarifying the organizational micro-foundations through which GAI enables sustainable and service-oriented value creation, without treating digital servitization as an explicit focal construct (Verhoef et al., 2021). Furthermore, the moderating role of DL is employed in order to confirm its significance in creating a culture in which AI integration can take place. A leader with a strong vision concerning digitalization, as well as one who supports innovation in the technological sector, is important in increasing the intentions of employees toward embracing the latest technologies.
Practical implicationsThe study provides specific management inspiration and practical paths for organizations that hope to achieve long-term performance improvement in digital transformation (Basha & Rodríguez, 2025). First, the study shows that building an open, trust-oriented, and technology-inclusive KS culture is the core foundation for unleashing the potential of GAI. In practical operations, companies can break information silos, promote knowledge flow and experience interaction among employees, and improve organizational knowledge agility and innovation diffusion efficiency by setting up AI knowledge co-creation days, building internal collaboration platforms, and introducing AI use case sharing mechanisms (Radanliev, 2025). Specifically, in the technology development context, it can be ensured that the efficient KS process not only enables compressing the implementation cycle to the application of technology but also enables the organization to be more responsive or adaptable in environments characterized by uncertainties. The study focuses on the crucial role played by employees' attitudes toward the emergence of AI for the ease of acceptance of the technology and enhancement of SOP. For efficient management practices in the organization, it should help the employees enhance AI application capabilities through the organization of training camps for AI capacity growth, seminars for GAI application, focusing on less apprehension toward the application of the technology to better prepare them for engaging with it. The study also discusses establishing AI ethics and compliance guidelines for AI usage, including the transparency and protection of data, to set up a stable and secure organizational context for AI implementation, hence boosting employees' confidence toward technology and establishing a virtuous cycle of AI acceptance, positive employee attitudes, and continued organizational performances.
Research findings show that DL plays an essential part in AI adoption. High-level digital vision and guidance from leaders are not only capable of leading their organization to understand the value of AI technology but are also able to establish an environment in which technology adoption will become favorable. For instance, leaders have the capacity to enable their organization's workers to embrace the hard results of intelligent technology using the "AI-driven innovation" concept in the strategy context of their organization, known as the "Digital Leader Lecture Hall," in addition to planning AI projects within the organization's various departments. The high DL within this study is essential in fostering the enhancement of GAI A influence on workers' KS, in addition to their attitudes, to amplify the performance environment.
In addition to operational efficiency and leadership support, organizations must also develop solutions for various ethical and normative challenges associated with GAI A. Issues concerning algorithmic transparency, data privacy, accountability, and employee trust can decisively influence how attitudes toward and acceptance of AI technologies unfold in practice (Faraj et al., 2018). The results indicate that DL is important in assuaging ethical concerns by establishing open communication, setting the boundaries within which AI is used, and matching AI deployment to organizational values (Raisch & Krakowski, 2021). Employees are more willing to trust AI-based systems and more actively participate in KS and innovative processes when ethics are incorporated into the transformation roadmap process. Further, KS can also act as an ethical governing mechanism when the collective knowledge within a group concerning AI capabilities and limitations is centralized, thus ending potential uncertainties and misuse. Ethics being integrated within AI leadership and practices will ensure responsible adoption and SOP related to AI adoption (Dwivedi et al., 2021).
Such ethical and governance issues might be more visible within organizations with a significant level of operation complexity, such as organizations linked with manufacturing and logistics activities, where issues of accountability, transparency, and system trustworthiness become relevant (Radanliev, 2025). Such organizations require DL as an important factor in governing control of proper use and avoiding possible risks.
Based on the analysis outcome of the empirical model, this study has developed a comprehensive governance structure for the five essential constructs (GAI A, KS, AT AI, SOP, and DL). The proposed structure not only provides a paradigm for linkages among technology, knowledge, psychology, and performance, expressed as “technology-knowledge-psychology-performance,” but also provides a guideline for enterprises to implement AI strategies. For example, for technology constructs, enterprises can implement GAI for customer service automation, text generation, and composing business reports for better efficiency and time responses. For human resource constructs, enterprises can use AI for employee skills analysis and AI-based customized employee training. For environmental, organizational performance constructs, enterprises can use AI for resource optimization, better supply chain transparency, and better capacities for eco-friendly governance. For leadership mechanism constructs, enterprises can use AI for better cross-functional collaborative capacities and better leadership capacities for innovations via DL. Enterprises will, therefore, be able to accomplish a synergistic advancement on economic benefits, environmental accountability, and social values by entering into a collaborative partnership on these five fronts.
Limitations and future research directionsThis study has several limitations. The data were collected using a convenience sampling approach and relied on cross-sectional, self-reported measures; this limits the finding's generalizability and restricts strong causal interpretation. The research was conducted within a specific geographical and institutional context (China), comprising participants from a range of industries. Hence, a generalization had to be made regarding applicability in contexts different in terms of regulatory frameworks, cultures, and levels of advancement in technology. For further improvement in external validity, cross-region or cross-economy studies could be considered, such as a comparison study between China and South Korea, which have quite different levels of digital maturity, regulations, and digital governance structures. Cross-region studies would provide more insightful information regarding how various contextual influences affect individuals' perceptions, attitudes, and motivation levels toward the usage of digital technology.
Furthermore, the current study also considers a restricted number of control variables. However, other variables that may also influence the study, such as cultural moderators or other organizational aspects, remain underexamined. Casting a broader conceptual framework that considers other psychological aspects of user attitude, such as user confidence, threat, or technology-related anxiety, could offer deep insight into the GAI A process on an organizational front.
The research provides empirical insights into the significance of GAI A in SOP; however, the study did not address the issues related to the processing of digital servitization or transformation of the business model. In this regard, future studies could advance the theoretical model using the study design as longitudinal or experimental studies, or using objective measures, among other variables, or through exploring the idea of digital transformation in other approaches. These would indeed help in understanding the significance of the impact of an organization’s sustainability in the long-term and ever-changing dimensions using GAI.
Ethics statementThis research involved human participants and complied with the ethical requirements in conducting research among human subjects. Participating in the research was entirely voluntary. Before the respondents participated in the research, they were informed about the purpose of the research and gave informed consent. The survey was administered anonymously, and no personally identifiable information was collected. All the collected data were handled confidentially and used solely for academic research purposes. According to applicable institutional and national regulations, formal ethical approval was not required for this type of anonymous survey-based study.
CRediT authorship contribution statementJiaming Hu: Writing – original draft, Methodology, Investigation, Data curation, Conceptualization. Yannan Li: Writing – review & editing, Validation, Supervision, Software.
The authors would like to thank the anonymous reviewers and editors for their valuable comments and suggestions that helped improve the quality of this paper.
