Next Article in Journal
Construction and Variation Analysis of Comprehensive Climate Indicators for Winter Wheat in Beijing–Tianjin–Hebei Region, China
Previous Article in Journal
Promoting Sustainable Manufacturing Growth Through Foreign Direct Investment and Foreign Trade: Evidence from China’s 28 Sub-Industries
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 17
Issue 20
10.3390/su17209060
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

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

Digital Learning Empowering Sustainable Education: Evidence from the Determinants of Chinese College Students’ Knowledge Innovation Capability

by
Yan Huang
1,
Zhihui Zhang
2,
Bingqian Xu
1,
Xinyu Zhou
2,
Jiayu Zhai
2 and
Da Gao
3,*
1
School of Management, Wuhan Institute of Technology, Wuhan 430205, China
2
School of Marxism, Wuhan Institute of Technology, Wuhan 430205, China
3
School of Law and Business School, Wuhan Institute of Technology, Wuhan 430205, China
*
Author to whom correspondence should be addressed.
Sustainability 2025, 17(20), 9060; https://doi.org/10.3390/su17209060 (registering DOI)
Submission received: 3 September 2025 / Revised: 24 September 2025 / Accepted: 10 October 2025 / Published: 13 October 2025
(This article belongs to the Section Sustainable Education and Approaches)
Browse Figures
Versions Notes

Abstract

With the rapid advancement of artificial intelligence technology, the role of Artificial Intelligence Generated Content (AIGC) applications within digital learning communities has become increasingly significant. Enhancing the level of knowledge innovation through the integration of human and artificial intelligence has emerged as a critical issue. Grounded in social cognitive theory, this study utilizes a sample of 407 Super Star Learn community learners as a case study. It applies the Fuzzy Set Qualitative Comparative Analysis (fsQCA) method to investigate the synergistic effects of technological environment, cultural context, and individual cognitive factors in promoting learners’ knowledge innovation capabilities. The results show the following: (1) No single condition constitutes a prerequisite for learners to achieve high-level knowledge innovation when acting in isolation. However, enhancing technical capabilities has a relatively universal impact on promoting learners to achieve these results. (2) The multiple concurrency of the technological environment, cultural environment, and individual cognitive conditions has generated multiple configuration patterns that promote knowledge innovation, indicating that the paths leading to learners’ high-level innovation exhibit the characteristic of numerous concurrency. Therefore, it is suggested that digital learning communities actively explore new paths for sustainable knowledge innovation and development driven by generative artificial intelligence technology, thereby injecting sustainable impetus into the development and innovation process of learners, contributing to the goals of sustainable education.

1. Introduction

With the emergence of a large number of virtual communities in the network environment and their application in the field of education, digital learning communities have come into being [1]. Digital learning community breaks through the space–time and organizational limitations of traditional knowledge exchange, diversifies knowledge, and provides a new platform and development space for knowledge innovation [2]. As a critical aspect of knowledge innovation research, learner knowledge innovation in a digital learning community involves the dynamic evolution of new knowledge through processes such as knowledge exchange, diffusion, absorption, and integration within the community [3]. Nowadays, the gradual integration of Artificial Intelligence Generated Content (AIGC) into digital learning communities provides a new technical toolset for learner knowledge innovation. However, it also presents learners with practical challenges [4]. Promoting the deep application of generative artificial intelligence and improving the level of learner knowledge innovation is key to overcoming the challenges of modernizing knowledge innovation in digital learning communities toward sustainable development.
Generative artificial intelligence has quickly processed massive data in the digital learning community, helping learners master richer and valuable information [5]. More importantly, with the intervention of generative artificial intelligence, the knowledge service mode and knowledge emergence mode of digital learning communities are gradually reshaping, which provides a key power engine for human intelligence knowledge transformation and collaborative innovation [6]. All of these indicate that the knowledge innovation of the digital learning community has undergone significant changes, and generative artificial intelligence has become a key partner in learner knowledge innovation. However, excessive dependence on intelligent tools may also cause learners to develop a certain degree of dependence and inertia [7]. Although it can quickly provide a large amount of rich learning information, the accuracy of this information cannot be guaranteed.
To address the issue of learner knowledge innovation in digital learning communities, the existing studies have conducted numerous practical explorations. Some scholars have put forward the influence of organizational culture’s virtual knowledge innovation environment on learner knowledge innovation [8] and the relationship between learner individual factors and knowledge innovation [9]. However, the current research on learner knowledge innovation is still in its infancy, focusing on the separate effects such as the technical environment [10], cultural environment of the digital learning community [11], and the individual level of learners [12]. Existing studies lack empirical evidence on the linkage matching effect between multiple factors, such as the technical environment, cultural environment, and individual learners, that are behind the differences in learner knowledge innovation. Therefore, it needs to deeply explore the configuration that drives learners’ high-level knowledge innovation, and effectively distinguish the core and edge conditions that affect learners’ knowledge innovation.
The influence of different factors on learner knowledge innovation in a digital learning community is not independent, different combinations could be formed through collaborative matching to influence learner knowledge innovation. Therefore, using the configuration perspective to carry out research is conducive to deepening the understanding of the complex mechanisms behind the knowledge innovation of learners in the digital learning community [13]. Based on the practical background of knowledge innovation in the digital learning community in the context of human–intelligence collaboration, this study uses the Fuzzy Set Qualitative Comparative Analysis (fsQCA) method to empirically explore the impact of three factors of technological environment, cultural environment, and individual learners on learner knowledge innovation, and finds the driving path to improve the level of learner knowledge innovation.
This study attempts to answer the following two specific questions: (1) What conditional configurations drive learner knowledge innovation in multiple concurrency? (2) Which conditions are more critical in improving knowledge innovation among learners? This study uses the social cognitive theory model to develop a research framework influencing knowledge innovation. Drawing on the practical context of knowledge innovation within the digital learning community of Super Star Learn—a prominent IT exchange platform in China—this study unveils the conditions and mechanisms that underpin variations in the level of learner knowledge innovation. It deepens our rational understanding of the sustainable development path and driving mechanisms of knowledge innovation in digital learning communities, and provides new evidence for research in learner knowledge, innovation, and human intelligence for long-term sustainability. It also provides empirical references for enriching and refining the theoretical framework of sustainable education, thereby aiding in the exploration of development strategies that better align with contemporary demands in areas such as curriculum design and pedagogical innovation.

2. Literature Review and Research Framework

2.1. Research on the Influencing Factors of Learner Knowledge Innovation in a Digital Learning Community

Bandura [14] proposed a social cognitive theory that incorporates the interaction of three key factors: individual, environment, and behavior, which the three-dimensional interaction model examines the dynamic interaction among them. They are independent of each other but can also influence one another. For example, individual cognitive factors, as mediators of external stimuli and responses, determine the behavior of individuals in the face of environmental stimuli. The environment, as the object or realistic condition of behavior, determines the direction and intensity of behavior [14]. Within the framework of the social cognitive theory model, research on learner knowledge innovation in digital learning communities can be carried out from three dimensions: individual, environmental, and behavioral factors, stimulating learners’ continuous high-level knowledge innovation.
The digital learning community provides learners with a convenient space for cooperation, allowing learners to exchange and innovate value sustainably [15]. Given the influencing factors of learners’ knowledge innovation in digital learning communities, a series of studies have focused on the community environment and individual learners. For example, some scholars have pointed out that the individual factors of virtual community learners (self-efficacy, creativity, perceived comparative advantage, compatibility, etc.) and virtual community environmental factors (reciprocity, norms, trust, community incentives, etc.) have a significant impact on the knowledge innovation behavior of community learners [16,17,18].
As for the influence of environmental factors, some studies have divided the virtual knowledge innovation environment into two categories: the knowledge innovation environment constructed by technology and organizational culture [8,19]. On the one hand, from the technology of learner knowledge innovation, the academic community has begun to study the possibility of generative artificial intelligence deeply participating in knowledge innovation or even independently conducting knowledge innovation with the rapid development of AIGC [20]. For example, some studies suggest that generative artificial intelligence has the technical characteristics of creativity. It can extract new output content like human beings based on existing information, which pushes the mode of knowledge production to a new height [21,22]. Some studies suggest that generative artificial intelligence is a reflective active system, which enables it to simulate human cognitive processes, accelerating at the knowledge level [23,24]. The dual labor subject, formed by the combination of generative artificial intelligence and the creative workers, exhibits a new feature of non-linear knowledge innovation level [25].
On the other hand, from the perspective of organizational culture, some scholars, based on the social capital theory, put forward that environmental factors such as incentive mechanisms, trust relationship, and group norms are conducive to stimulating learners’ knowledge innovation behavior [26,27,28]. Specific to the impact of individual factors, Parhamnia et al. [29] suggest that the learner’s self-efficacy and expected outcome significantly affect the virtual learning community knowledge innovation; there are also studies that the differences in the growth environment, background experience and personal ability of different knowledge subjects lead to their differences in social attributes such as knowledge stock, knowledge level and knowledge communication, which are called knowledge distance. The closer the knowledge distance between different subjects is, the easier it is for them to break through knowledge barriers and form a consensus, promoting the generation of knowledge innovation achievements in virtual learning communities [30,31].

2.2. Construction of Learner Knowledge Innovation Behavior Model in Digital Learning Community

Although existing research has discussed the differentiation of learner knowledge innovation paths in digital learning communities, there are still few studies that can be used to explain the differentiation path of knowledge innovation in digital learning communities in the context of human–intelligence interaction. The existing research has the following limitations: First, although it provides a sufficient explanation for the knowledge innovation of learners in the digital learning community, it is insufficient to provide strong theoretical support for the differentiated path selection to improve the level of learner knowledge innovation. Second, the improvement of the learner’s knowledge innovation level is the result of the joint action of multiple antecedent conditions, rather than the independent role of a single condition. The existing research has unified the symmetric relationship between independent and dependent variables, restricting the path selection to improve the knowledge innovation level of learners in digital Learning communities. Third, in digital learning communities, learner knowledge innovation emerges from specific configurations of antecedent conditions and their logical relationships with outcome variables. The critical question is which configurations of antecedent conditions lead to positive outcomes. These conditions that lead to high-level knowledge innovation of learners may not be the same as those.
The existing research has not yet adequately addressed the complexity of the causal relationship between learner knowledge innovation and digital learning communities. Given these shortcomings, this study employs the fsQCA method to explore the synergistic matching effect and interactive relationship between the technological environment, cultural environment, and learner individual factors on knowledge innovation in digital learning communities. Based on the analysis framework of social cognition theory, this study combines generative artificial intelligence with learner knowledge innovation practices. It constructs a configuration effect model that affects learner knowledge innovation in digital Learning communities. The construction of this model not only provides a theoretical tool for studying knowledge innovation in digital learning communities, but also offers a reference analytical framework for enhancing learners’ innovation capabilities through multi-factor synergy in sustainable education, as shown in Figure 1.
(1)
Technical environment. Specifically, it includes two secondary conditions: creativity and reflexivity of generative artificial intelligence. In the interaction between learners and technology, the characteristics of technology would influence learners’ adoption and application of technology [32]. Based on the transaction cost theory, the cultivation of knowledge productivity requires the output of learners’ innovative knowledge. The generative AI can not only realize the accurate integration and utilization of resources, but also help learners quickly obtain relevant resources at a low cost for knowledge innovation [33]. It can also learn from large-scale data resources and create new knowledge and content, which has a positive impact on promoting learner knowledge innovation [15]. The creativity of AIGC is a feature that supports the creative transformation and innovative development of knowledge production. AIGC’s creativity stems from the technological characteristics driven by generative AI algorithms’ generation mechanisms and multimodal output capabilities. This capability enables AI to generate novel, diverse, learning-oriented content based on existing knowledge repositories, thereby driving the creative transformation and innovative development of knowledge production [34]. This feature has spawned a new knowledge production model, producing additional knowledge and effectively promoting learner knowledge innovation [35]. Meanwhile, generative AI demonstrates technical capabilities that mimic human reflective processes, including analyzing input information, learning from interactions, and generating adaptive responses based on established patterns. First, the system records user revision suggestions and query trajectories through feedback loop algorithms, then adjusts subsequent output logic via reinforcement learning. Second, it further decomposes answer derivation processes using logical chains stored in pre-trained models. Ultimately, by analyzing user behavior data, the system can identify deficiencies in its own outputs and optimize response strategies for similar queries. This technical capability, determined by platform architecture rather than individual cognitive capacity, constitutes another core element driving collaborative knowledge innovation between AI and users [36].
(2)
Cultural environment. Specifically, it includes two secondary conditions: a trust relationship and an incentive mechanism. A trust relationship is an important psychological driving factor for learners to exchange information and share knowledge in a digital learning community, which is conducive to promoting learners’ knowledge contribution and innovation [15]. The trust relationship in the digital learning community encompasses two levels: member trust and community trust [37,38]. When learners have a sense of trust in community members, they are willing to take altruistic teaching assistant knowledge contribution behavior to promote knowledge innovation [39]. According to the trust transfer theory, the trust between community members could be derived from their trust in the community itself and the content of the community. When they believe that their knowledge innovation behavior could contribute to the community’s security, they will be more willing to participate in community activities, take the initiative to strengthen interaction with other learners, and promote the overall level of knowledge innovation in the community [15]. Expectancy theory of motivation holds that an individual’s behavior is driven by the reward expected from the organization, which includes material rewards, identity within the organization, or improved interpersonal relationships [40]. The incentive mechanism established by the digital learning community can effectively stimulate learners’ curiosity, self-challenge, and the spirit of exploration, thereby innovating and expanding existing knowledge sources.
(3)
Individual cognition. Specifically, it includes three secondary conditions: self-efficacy, expected outcome, and knowledge distance. Self-efficacy refers to an individual’s perception and judgment about their ability to complete a specific activity. In the digital learning community, learners with high self-efficacy typically possess comprehensive abilities. They are more confident that they can overcome the difficulties in the process of knowledge innovation, contribute more valuable new ideas and new methods, and promote knowledge innovation [41]. The expected outcome is the learner’s cognitive judgment of the possible results brought by their knowledge innovation behavior. According to expectancy theory, whether the individual’s expectation of the behavior result is positive or negative is an essential factor in the occurrence and maintenance of behavior [42]. A positive expected outcome can contribute to learners’ knowledge and informed decisions about innovative behavior. For example, learners will have a positive attitude towards knowledge innovation if they gain external or internal benefits, including financial rewards, promotion, or self-satisfaction, as well as social recognition [43]; on the contrary, a negative expected outcome will inhibit learners’ behavioral decision-making. Knowledge distance is the gap or degree of dissimilarity between learners’ knowledge level or knowledge content [29]. It is an essential antecedent variable that affects knowledge innovation. According to the knowledge potential theory, the small knowledge distance between the two sides means that the knowledge potential difference between each other in a teaching assistant’s field is slight, which is conducive to the absorption of the received knowledge by learners, providing favorable conditions for knowledge evolution and realizing knowledge co-creation [30,44].
Based on the above analysis, among the seven secondary conditions included in the technical environment–cultural environment–individual cognition analysis framework, creativity, reflexivity, trust relationship, and incentive mechanism belong to the objective environmental factors of learner knowledge innovation, and self-efficacy, knowledge distance, and expected outcome belong to the individual cognitive factors of learner knowledge innovation.

3. Research Design

3.1. Sample Selection and Data Collection

This study selects the Super Star Learn digital learning community as a research platform. The platform utilizes generative artificial intelligence technology to develop AI teaching assistants, such as “Socrates Star”, providing a practical platform for researching human–AI collaborative innovation in digital learning communities. This study combines measurement questionnaires of relevant variables from existing literature with a survey questionnaire designed for knowledge innovation practices in the Super Star Learn community. Since some existing questionnaires are originally written in English, this study invites two researchers with strong English proficiency to translate these original questionnaires using the ‘translation-back translation’ method. The researchers then revised the questionnaires based on the specific context of this study to form the initial measurement tools for this research. After the initial questionnaire is developed, three experts in the fields of human–machine collaboration and knowledge innovation are invited to review and refine the questionnaire.
Additionally, 50 learners who had used SuperStar Learning Pass (https://v8.chaoxing.com/) and its AI teaching assistant function are invited to participate in a pre-test, and the questionnaire is revised again. The formal questionnaire is disseminated through multiple channels, including Questionnaire Star, MicroBlog, WeChat groups, QQ groups, TikTok, and Rednote, to collect sample data. Between 1 September and 23 December 2024, a total of 578 questionnaires are collected. After excluding invalid questionnaires with answers that are contradictory to other options, exhibit obvious patterns, or fail to meet the five-minute minimum completion time, the number of valid questionnaires is 407.

3.2. Variables and Their Measurement

This study primarily draws on mature scales from domestic and foreign literature for variable measurement. It combines the results of expert recommendations and pre-tests to refine the measurement, ensuring its accuracy. First of all, for the measurement of the technical and cultural environment of the digital learning community, this study mainly refers to the research of Acar et al. [45], Lucas et al. [46] and Kmieciak [47] by using two items to measure creativity; two items to measure reflexivity; two items measure the trust relationship, including one item measuring community trust and one item measuring member trust—two item measurement incentive mechanism. Secondly, for the measurement of learners’ cognition, this study mainly refers to the items in Cheung [48] and other studies, using two items to measure self-efficacy, two items to measure expected outcome, and two items to measure knowledge distance. Finally, this study primarily draws on the research of Yli-Renko et al. [49] and uses three items to measure the knowledge innovation behavior of learners in the digital learning community. All items were scored using a 7-point Likert Scale, ranging from “very disagree” to “very agree,” which was recorded as 1–7 points. The specific information of all items is shown in Table 1.

3.3. Reliability and Validity Analysis and Homologous Variance Test

This study examines the reliability of the data before testing the relationship between variables, ensuring the validity of the research conclusions. The results are presented in Table 1 and Table 2. In this study, composite reliability (CR) is used to test the reliability of the data. The Cronbach’s α value of all variables was above 0.7, and CR is above 0.8, indicating that the measurement of variables in this study had good internal consistency. Validity tests include content convergent and discriminant validity. The content validity test is supported because the measurement tool used in this study is based on a mature scale, revised by experts, and developed after the pre-test, ensuring good content validity. The convergent validity test shows that Confirmatory Factor Analysis (CFA) is used to calculate and determine the Average Variance Extracted (AVE).
The CFA results show that the overall fitting index of the measurement model is good. x 2 = 576.832, d f = 265, x 2 / d f = 2.177, IFI = 0.914, TLI = 0.910, CFI = 0.913. The factor loading values of all items in this study were greater than 0.7, and the AVE is greater than 0.5, indicating that the convergent validity of the measurement model met the requirements. In the discriminant validity test, this study compares the correlation coefficient between the square root of each latent variable and each latent variables. The results show that the square root of each latent variable AVE is greater than the correlation coefficient between each latent variable, indicating that the measurement model has good discriminant validity. In addition, considering that most items are completed through the learner’s self-assessment, this study also used SPSS 27.0.1 software for Harman’s single-factor test to assess whether there is a significant standard method bias problem. The results show that the maximum component factor is 36.424%, which is lower than the baseline of 40%, indicating that the common method bias does not have a significant impact on this study.

3.4. Variable Calibration

In fsQCA analysis, each antecedent condition and outcome variable is considered as an independent set, and each case in these sets has a membership score. Following Ragin [50], we defined three anchors required for calibration surveys or continuous datasets to establish fuzzy set values that determine the membership degree of each score. This method, applicable to fuzzy set conversion of constant variables, effectively distinguishes high, medium, and low levels of conditional variables. Using the 0.95, 0.50, and 0.05 percentiles is an established practice in fsQCA research [51,52]. This method defines thresholds for full membership and full non-membership based on the extreme values of the sample distribution. It effectively identifies the clearest cases of fully belonging and fully not belonging within a set, which is crucial for meaningful set-theoretic analysis. Therefore, this study employs the direct calibration method to set the calibration criteria for full membership at the 0.95 percentile, the intersection point at the 0.5 percentile, and full non-membership at the 0.05 percentile for creativity, reflectivity, trust relationships, incentive mechanisms, self-efficacy, outcome expectations, and knowledge gap.

3.5. Fuzzy-Set Qualitative Comparative Analysis

This study aims to analyze the multiple driving mechanisms behind learner knowledge innovation in the digital learning community from a configurational perspective. It intends to employ the Qualitative Comparative Analysis (QCA) method to conduct the research. Compared with the traditional statistical method based on the binary relationship, the advantages of QCA are mainly reflected in the following. In the multi-path analysis of the improvement of learners’ knowledge innovation level in the digital learning community, there are multiple interactions among creativity, trust relationship, self-efficacy, and other factors. They not only influence each other but also affect the overall level of learners’ knowledge innovation. In addition, the traditional regression methods cannot deal well with this multiple interaction problem. QCA analysis can explore the differentiated driving mechanisms of learner knowledge innovation in digital learning communities from the perspective of holistic relationships. Therefore, this study employs the QCA method to investigate the complex mechanism of multiple factors on the level of learner knowledge innovation.
The diversity of learners’ high-level knowledge innovation paths in the digital learning community shows that there may be an equivalent causal chain among the technological environment, cultural environment, and learner individual cognition that jointly drives learners’ high-level knowledge innovation paths. Although traditional multivariate statistical methods can describe the influence of independent variables on dependent variables through mediating and moderating variables, they typically use the substitution relationship or cumulative relationship of independent variables to explain the variation in dependent variables, rather than the complete equivalence relationship.
Regarding specific analytical techniques, the QCA methodology encompasses three types: Multi-Valued Set Qualitative Comparative Analysis (mvQCA), Exact Set Qualitative Comparative Analysis (csQCA), and Fuzzy Set Qualitative Comparative Analysis (fsQCA). Constructing conditional variables based on membership degrees overcomes the limitations of classification-based scoring inherent in earlier methods such as csQCA and mvQCA [53]. This method integrates predetermined conditions through Boolean logic operations, analyzing cases after considering alternative scenarios for the target analysis [54]. This study employs the fsQCA method to deeply explore the conditional configurations and causal mechanisms of learners’ sustainable knowledge innovation, thereby contributing to the relevant research field [50,55]. This contribution is achievable only through fsQCA, as it delivers more comprehensive analytical outcomes compared to traditional quantitative statistical methods that typically provide singular interpretations for specific dependent variables [55]. Consequently, fsQCA demonstrates significant advantages over the other two categories of QCA techniques. For these reasons, fsQCA was selected as the specific analytical technique for this study.

4. Data Analysis and Empirical Results

4.1. Analysis of Necessary Conditions

Prior to performing conditional configuration analysis, this study individually examined the “necessity” of each condition. In line with mainstream QCA research, it first evaluated whether a single condition, including its non-configurational form, constituted a necessary condition for user knowledge innovation. Necessity analysis assesses whether each antecedent condition is indispensable for the outcome condition, defined as a causal context that persists when the outcome occurs. The consistency score for necessity analysis should exceed 0.90, with coverage exceeding 0.5 [56,57]. Table 3 presents the test results of the necessary condition analysis for high-level and non-user high-level knowledge innovation using the fsQCA 4.1 software. Findings show that the consistency of all conditions falls below 0.9, indicating that none of the seven conditions examined in this study are necessary for influencing high-level or non-user high-level knowledge innovation.

4.2. Conditional Configuration Analysis

The sufficiency analysis of conditional configuration involves examining the combination of the seven antecedent conditions to determine the impact of their constituent elemental configurations on the generation of knowledge innovations. The consistency level is also used to measure the degree to which the configuration explains the result variable, but the consistency level for determining sufficiency should not be less than 0.75. According to the specific research situation, existing studies have adopted different consistency thresholds, such as 0.8 [58], 0.85 [59], among others. The frequency threshold should be determined based on the sample size, and the distribution of cases in the truth table, as well as the researchers’ familiarity with the observed cases, should also be considered in the specific research. For small and medium samples, the frequency threshold is 1; for large samples, the frequency threshold should be greater than 1. The finalized consistency threshold for this study is 0.80, and the frequency threshold is 2, which finally covers 407 samples.
Currently, research on knowledge innovation in digital Learning communities within an intelligent collaborative context is still in the exploratory stage. Until November 2022, the emergence of generative artificial intelligence, represented by ChatGPT 3.5, had not yet sparked significant academic research on knowledge innovation in digital Learning communities within an intelligent collaborative context. Along with the widespread social application of generative AI, the natural language large model it exhibits has already possessed some features oriented to general AI [60], such as the emergence of unique human-like thinking as well as the ability of learning, creativity, and reflexivity, etc., which improves the level of knowledge innovation in the digital learning community. From the practice of knowledge innovation in the ‘Super Star Learn’ community, it can be seen that with the help of generative artificial intelligence technology, the ‘Super Star Learn’ community has improved the learning efficiency and knowledge innovation level of learners. Therefore, this study chooses ‘exist’ for the technical environment. Through the design of incentive mechanisms, learner interaction mechanisms, and real-name authentication, ‘Super Star Learn’ community stimulates learners’ motivation to learn and enthusiasm to participate, enhances the sense of trust among learners, and improves the level of learners’ knowledge innovation. In this study, the incentive mechanism and trust relationship variables are taken as the ‘existence’ conditions. For the self-efficacy variable, learners with low self-efficacy tend to choose ‘presence or lack’ for self-efficacy because they lack confidence in their own knowledge and prefer to retain their ideas rather than share them in knowledge contribution, which reduces their willingness to participate in knowledge innovation activities [61]. The individual cognitive conditions of creativity and knowledge distance were both selected as ‘present or lacking’. The results of the group analysis of the seven conditions of this study on high-level knowledge innovation are shown in Table 4.
The four groupings presented in Table 4 can be considered a combination of sufficient conditions for high-level knowledge innovation among learners. In this study, fsQCA analysis was used to identify four configurations that achieve high levels of knowledge innovation among learners, and the consistency of each configuration is greater than 0.75, indicating that each configuration is a sufficient condition for high levels of knowledge innovation among learners. The consistency of the overall solution for these four configurations is 0.961, indicating that approximately 96.1% of cases satisfying these four configurations involve high-level knowledge innovation. The coverage of the overall solution is 0.652, indicating that these four configurations can explain about 65.2% of cases involving high-level knowledge innovation among learners. Based on the core conditions of these four configurations, three typical patterns for realizing high-level learner knowledge innovation can be derived.
Configuration 1 suggests that learners’ possession of high self-efficacy and expected outcome is a unique pathway to enhance the level of knowledge innovation. It also implies that the existence of learner self-efficacy (individual cognition) and expected outcome (individual cognition) as the core conditions is crucial for learners’ high level of knowledge innovation, and that these two elements can together constitute a sufficient condition to explain the generation of learners’ high level of knowledge innovation results. Therefore, this study names this pattern as ‘individual cognitive type’, which also indicates that high self-efficacy and expected outcome can effectively break the constraints of knowledge innovation by objective conditions such as technological and cultural environments. The consistency of configuration 1 is 0.983, and the original coverage and unique coverage are 0.451 and 0.201, respectively. This indicates that this path can explain approximately 44.6% of the learners’ knowledge innovation cases, and about 20.1% of the knowledge innovation cases can only be explained by this path.
This study examines three exemplary cases of high-performing knowledge innovators from Configuration 1, focusing on two key research questions: “Why are self-efficacy and outcome expectations crucial for knowledge innovation?”, and “How can high self-efficacy and positive outcome expectations compensate for imperfect community environments?”. Through 30–40 min in-depth interviews, participants revealed three critical success factors: first, strong self-confidence that persists through challenges (I’m willing to try new things despite difficulties); second, anticipated rewards from peer recognition (Sharing ideas brings honors and respect); and third, future-oriented motivation (This hope drives my innovation). Notably, all participants reported generating valuable insights through peer collaboration even in suboptimal environments, demonstrating that robust individual cognitive factors can effectively overcome environmental limitations in digital learning communities.
In configuration 2, the existence of trust relationships and expected outcome conditions play a central role. In the knowledge innovation activities of learners in the digital learning community, high-trust relationships and high-expected outcome conditions can break through the limitations of technical environment conditions, enabling learners to achieve a high level of knowledge innovation. Considering that configuration two consists of the trust relationship (cultural environment) condition and the expected outcome (individual cognition) condition, this study refers to this driving path as ‘individual cognition–cultural environment type’. The consistency of configuration 2 is 0.971, and the original and unique coverage are 0.303 and 0.070, respectively. The results show that this path explains about 30.3% of the cases of learner knowledge innovation, and only about 7% of the cases of knowledge innovation can be explained by this path.
In configuration 3, self-efficacy plays a central role, and the existence of incentives, expected outcome, and knowledge distance plays a supporting role. The driving path is composed of four conditions: self-efficacy (individual cognition), expected outcome (individual cognition), knowledge distance (individual cognition), and incentive mechanism (cultural environment), which is why configuration three is also named ‘individual cognition–cultural environment type’ in this study. The consistency of configuration 3 is 0.986, and the original coverage and unique coverage are 0.397 and 0.151, respectively. This path accounts for approximately 39.7% of the cases, while only 15.1% of the cases can be attributed to this path.
The presence of technological creativity in configuration 4 plays a central role, while the presence of reflexivity, trust, relationship conditions, and incentive mechanisms complements this role. In the digital learning community with a high level of trust relationship and well-designed incentive mechanism, even if the learners’ self-efficacy and expectation of results in knowledge innovation activities are insufficient, with the support of technological environmental conditions, the learners can obtain valuable information and inspiration for knowledge innovation from new content and knowledge provided by generative AI, and to improve the level of knowledge innovation. The driving path consists of four conditions: creativity (technological environment), reflexivity (technological environment), trust relationship (cultural environment), and incentive mechanisms (cultural environment). This study refers to this configuration as ‘technological environment–cultural environment type’. The consistency of configuration 4 is 0.987, and the original coverage and unique coverage are 0.312 and 0.134, respectively. This path accounts for approximately 31.2% of the cases of knowledge innovation, while only 13.4% of the cases can be explained by this path.
In this study, three high-level knowledge innovation learners were selected as a typical case of ‘technological environment–cultural environment’ knowledge innovation configuration 3, and the following questions were asked of these three interviewees: “What do you think is the value of ‘AI teaching assistant’ for your knowledge innovation activities?”, “What valuable assistance do you think an ‘AI teaching assistant’ has provided to your knowledge innovation activities?”, “How do you think the atmosphere of trust within the community and the well-developed incentives have helped you?”, and “When your knowledge innovation activities were frustrated and you felt lost about your academic future, could an ‘AI teaching assistant’ provide you with some help and inspiration?”. Participants reported that the AI assistant served as both mentor and collaborator—“It expands my thinking through multi-perspective questions and reference cases”, while its self-correcting capability boosted innovation confidence. The community’s trusting atmosphere (“I feel safe sharing ideas without fear of ridicule”) and incentive system (badges, points) significantly enhanced engagement. Notably, during creative blocks, the AI provided both technical guidance and emotional support, with one participant noting: “When stuck coding, it offers solutions and even humor to reset my mindset.”. These findings demonstrate how integrated technological and cultural environments synergistically support innovation. This collaborative model that supports innovation fosters a long-term, stable, and highly effective learning environment for innovation among learners. It represents a crucial manifestation of how sustainable education enhances educational quality through dual empowerment from technology and culture.

4.3. Configuration Model Robustness Testing: Overall Solution XY Scatter Plot Analysis

To further validate the robustness of the configurational analysis results, this study constructed an XY scatter plot to depict the relationship between the overall solution and the outcome variable, in accordance with the methodological framework of fuzzy-set qualitative comparative analysis (fsQCA) [50], as shown in Figure 2. As a core diagnostic tool for assessing configurational adequacy in fsQCA, the XY scatter plot visually represents the association between antecedent configuration membership and outcome variable membership, thereby providing empirical support for evaluating solution consistency [62].
In Figure 2, the horizontal axis reflects overall solution membership, indicating the degree to which cases align with the four high-knowledge innovation pathways. In contrast, the vertical axis represents knowledge innovation behavior membership, capturing the extent of high-level knowledge innovation demonstrated by each case. The findings reveal that approximately 96% of the data points are clustered along the diagonal and within the upper-left quadrant, indicating strong alignment with the expected solution path. This concentration underscores the explanatory power of the identified configurations in accounting for high-level knowledge innovation, consistent with established fsQCA criteria for solution adequacy [63]. A small number of outlier cases in the lower-right quadrant display high configuration membership but low outcome membership, suggesting possible contextual or unobserved moderating factors. These deviant cases merit deeper investigation in future studies to identify underlying mechanisms. Overall, the XY scatter plot provides a clear visual validation of the configurational model’s robustness and coherence, offering intuitive and compelling supplementary evidence that reinforces the study’s conclusions.

5. Conclusions and Discussion

5.1. Conclusions

In this study, the fsQCA method is applied to analyze the condition configuration of 407 learners’ knowledge innovations in the Super Star Learn community as a case study, to explore the linkage effects of the technological, cultural environment, and individual cognitive factors on learner knowledge innovation, and to reveal the core conditions affecting learners knowledge innovation and the nature of their complex interactions. The study shows that learner knowledge innovation in digital learning communities is characterized by multiple concurrency and diverse concurrency. The specific findings are as follows.
In general, the technological environment, cultural environment, and individual cognitive factors cannot stand alone as necessary conditions for high-level knowledge innovation of learners in digital Learning communities, suggesting that any single factor can neither constitute an essential condition for high-level knowledge innovation of learners nor be a sufficient condition for triggering high-level knowledge innovation of learners. The innovation of learners’ high-level knowledge is the result of multiple factors. Based on the configuration perspective, this study uses the fsQCA method to discover four equivalent paths of high-level knowledge innovation of learners, which can be specifically categorized into three types, namely, the fitness model independently explained by individual cognitive elements of learners, elements of the cultural environment, and the fitness model jointly explained by aspects of the technological environment and the cultural environment. Behind the knowledge innovation of learners in the digital learning community is the linkage of multiple factors, the effective combination of which enhances the level of learner knowledge innovation through diverse concurrency.

5.2. Theoretical Contribution

Compared to other studies on learner knowledge innovation in digital Learning communities, the theoretical contributions of this study are as follows.
First, this study shifts the research focus on knowledge innovation in digital learning communities from a fragmented view—emphasizing solely technical support, organizational culture, or individual cognition—to a holistic perspective that underscores multi-factor synergy. Based on social cognitive theory and using the fsQCA approach, it reveals multiple configurational paths affecting knowledge innovation capability in digital environments, offering new empirical support for the theory. Findings show that no single factor is necessary for high-level knowledge innovation—each functions only within specific configurations of other factors. This aligns with social cognitive theory’s emphasis on causal interactivity. Thus, the study validates the theory’s applicability in digital learning communities, extends its explanatory scope, and provides a theoretical reference for future research.
Second, current research on knowledge innovation among learners in digital learning communities is primarily confined to traditional quantitative analytical methods, such as studies employing structural equation modeling [64]. Notably, Hossain [65] has used the fsQCA method to analyze the antecedents of data-driven innovation empirically. His research demonstrates that fsQCA possesses unique advantages in exploring causal mechanisms involving multi-condition interactions. This methodological approach shares common ground with the causal complexity among conditions for learner knowledge innovation examined in this study. Building upon this foundation, this research innovatively introduces the fsQCA method. This not only enriches the methodological toolkit for studying knowledge innovation within virtual academic communities and makes incremental contributions to the formation of user knowledge innovation theory, but also provides a holistic perspective for exploring the causal complexity among various conditions underlying user knowledge innovation.
Third, at present, studies have confirmed that the process of human–computer collaboration has a positive impact on the efficiency and effectiveness of knowledge innovation; however, these studies have not yet fully considered the revolutionary breakthroughs in knowledge innovation brought about by the rapid development of generative AI technology and incorporated this technological sophistication into the field of human–intelligence collaboration innovation as an antecedent condition to carry out an in-depth study. This study finds that generative AI technology can effectively alleviate or overcome the constraints imposed by learners’ individual cognitive conditions on their knowledge innovation in specific cultural environments, thereby enriching the research on the mechanism of technological sophistication in enhancing learners’ knowledge innovation levels.

5.3. Practical Implications

The findings of this study have the following practical implications for digital Learning communities and learners:
Initially, digital Learning communities should actively explore new development paths for sustainable knowledge innovation empowered by generative AI, thereby injecting sustained impetus into the growth of community learners and knowledge innovation. This exploratory process also serves as a crucial initiative for advancing sustainable education. By providing learners with continuous support for knowledge innovation, it helps sustainable education achieve its goal of cultivating talent equipped with lifelong learning and innovative capabilities. Among the four paths to successful knowledge innovation identified in this study, generative AI can effectively mitigate or overcome the negative impact of low self-efficacy and low expected outcome on knowledge innovation under specific conditions, which is particularly important for enhancing the level of knowledge innovation of learners in digital Learning communities. This reveals that the digital learning community should keep pace with the development of science and technology, absorb advanced generative artificial intelligence technology progressively, create AIGC tools with advanced technical characteristics for learners, make these brilliant AIGC tools become the partner of learners’ knowledge innovation, and work with learners to carry out valuable knowledge production activities, and continue to promote learners’ knowledge innovation. In addition, the digital learning community also needs to adhere to humanistic values, balancing the relationship between science and ethics, fairness, and efficiency in the process of integrating generative AI technology into the community, and realizing the sustainable development of the digital learning community empowered by generative AI technology.
Second, for learners, they should fully integrate into the cultural environment of the digital learning community, continuously improve their self-efficacy, gradually narrow the knowledge distance with other members of the community, and enhance the overall knowledge innovation level of the community. Such self-improvement and collaborative development behaviors among learners help propel sustainable education from concept to practice, achieving shared sustainable development for both individuals and the educational system. This study found that the synergistic linkage of learners’ self-efficacy, creativity, and knowledge distance can enhance their overall knowledge innovation level. This suggests that learners should strive for the following: Firstly, continuously improve the depth and breadth of their knowledge learning, especially strengthen the learning and adaptation of generative AI applications, enrich the knowledge structure, improve the depth of thought, and continuously narrow the knowledge distance with peers, so that they can fully appreciate the value and significance of self-improvement, and transform this positive feedback into confidence in knowledge innovation. Secondly, the multiple incentives designed by the digital learning community can allow learners to perceive the rewards of engaging in knowledge innovation activities, thus enhancing their expected outcomes of participating in community knowledge innovation and injecting a strong impetus for community knowledge integration and innovation. Therefore, learners should be fully integrated into the cultural atmosphere created by the digital learning community to form positive feedback that sustains innovation [66].

5.4. Shortcomings and Prospects of the Research

This study has some limitations and shortcomings that warrant further research in the future.
First, constrained by the completeness and availability of text and data, this study examined only the configurations of 407 learners within a single community, “Super Star Learn”. This limitation may restrict the generalizability of the findings. Future research should consider conducting more in-depth investigations, such as case studies, large-scale data analysis, and validation of these findings across more diverse community contexts. Such approaches would complement this study and enhance the generalizability of knowledge innovation research conclusions. At the same time, it can provide richer empirical evidence for the implementation of sustainable education across diverse contexts, enabling the development of more adaptive strategies tailored to the unique characteristics of different communities. This will further promote the diversified advancement of sustainable education.
Second, it has been more than three decades since Rogers [67] first defined the concept of knowledge innovation, which is more widely recognized in the academic community. Still, there have been fewer studies on measuring knowledge innovation activities in virtual communities. Most of them focus on measuring the performance of knowledge innovation. For example, Frenz [68] measured the performance of knowledge innovation in two dimensions: innovation of knowledge sources and the development of new problems and new ideas. In this study, the measurement analysis of knowledge innovation activities in digital learning community learners was conducted from a group perspective; however, the research data primarily originated from learners’ self-reports, which introduces the possibility of subjective statistical bias. Future research can consider utilizing web crawler technology to collect objective data, such as the text of learners’ knowledge innovation-related activities. This data can then be analyzed using text analysis and other technologies to conduct a more comprehensive and accurate evaluation of learners’ knowledge innovation activities in digital Learning communities.
Third, relative to the existing literature, although this study has made some enrichments and expansions of the conditional variables in the knowledge innovation grouping analysis of digital Learning communities, there are still some topics that deserve in-depth research due to the limitations of space. For example, nowadays, along with the rapid development of generative artificial intelligence technology, AIGC’s content comprehension ability, autonomous thinking ability, and communication ability have been greatly enhanced. Future research on human intelligence, collaboration, and knowledge innovation is bound to yield revolutionary breakthroughs. This study is unable to provide a comprehensive and accurate prediction of the entire picture of knowledge innovation in the context of human–intelligence collaboration, which warrants in-depth exploration in the future. In addition, this study can also compare the configurations of learner knowledge innovation among different types of digital Learning communities, the configurations of knowledge innovation between leading and non-leading learners, and the configurations of knowledge innovation among the same learner in different digital Learning communities.

Author Contributions

Conceptualization, Y.H., B.X., X.Z. and D.G.; Methodology, Y.H., Z.Z. and J.Z.; Validation, X.Z.; Formal analysis, Y.H., Z.Z., B.X. and J.Z.; Investigation, Z.Z., B.X. and X.Z.; Resources, Z.Z., B.X., J.Z. and D.G.; Data curation, Y.H. and X.Z.; Writing—original draft, Y.H. and Z.Z.; Writing—review & editing, D.G.; Visualization, J.Z.; Supervision, D.G.; Project administration, B.X., J.Z. and D.G.; Funding acquisition, D.G. All authors have read and agreed to the published version of the manuscript.

Funding

This study is a phase-specific research outcome of the National Social Science Fund Project 2024, number (24BKS134).

Institutional Review Board Statement

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the Ethics Committee (EC) at Wuhan Institute of Technology (Protocol Number: 25Y00217, Approval Date: 15 July 2025).

Informed Consent Statement

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

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Daniel, B.; Schwier, R.; McCalla, G. Social capital in virtual learning communities and distributed communities of practice. Can. J. Learn. Technol./Rev. Can. l’Apprentissage Technol. 2003, 29. [Google Scholar]
  2. Burtch, G.; Lee, D.; Chen, Z. The consequences of generative AI for online knowledge communities. Sci. Rep. 2024, 14, 10413. [Google Scholar] [CrossRef]
  3. Huang, X.; Li, H.; Huang, L.; Jiang, T. Research on the development and innovation of online education based on digital knowledge sharing community. BMC Psychol. 2023, 11, 295. [Google Scholar] [CrossRef]
  4. Chen, Y.; Wu, S.; Chen, J.; Wu, X. The Dual Impact of AIGC Technology on College Students’ Learning Efficiency and Motivation: An Interview-Based Study. Int. J. New Dev. Educ. 2025, 7, 56–64. [Google Scholar] [CrossRef]
  5. Yu, H.; Guo, Y. Generative artificial intelligence empowers educational reform: Current status, issues, and prospects. Front. Educ. 2023, 8, 1183162. [Google Scholar] [CrossRef]
  6. Bond, M.; Khosravi, H.; De Laat, M.; Bergdahl, N.; Negrea, V.; Oxley, E.; Siemens, G. A meta-systematic review of artificial intelligence in higher education: A call for increased ethics, collaboration, and rigour. Int. J. Educ. Technol. High. Educ. 2024, 21, 4. [Google Scholar] [CrossRef]
  7. Naseer, A.; Ahmad, N.R.; Chishti, M.A. Psychological Impacts of AI Dependence: Assessing the Cognitive and Emotional Costs of Intelligent Systems in Daily Life. Rev. Appl. Manag. Soc. Sci. 2025, 8, 291–307. [Google Scholar] [CrossRef]
  8. Li, N.; Zhang, X.; Limniou, M. A country’s national culture affects virtual learning environment adoption in higher education: A systematic review (2001–2020). Interact. Learn. Environ. 2023, 31, 4407–4425. [Google Scholar] [CrossRef]
  9. Chou, S.W. Why do members contribute knowledge to online communities? Online Inf. Rev. 2010, 34, 829–854. [Google Scholar] [CrossRef]
  10. Oke, A.; Fernandes, F.A.P. Innovations in teaching and learning: Exploring the perceptions of the education sector on the 4th industrial revolution (4IR). J. Open Innov. Technol. Mark. Complex. 2020, 6, 31. [Google Scholar] [CrossRef]
  11. Kappel, M.; Judge, M.; Ng, C.Y.; Ntebutse, J.G.; Williamson-Leadley, S.; Yildiz, M.N. Learning in digital environments: A model for cross-cultural alignment. Educ. Technol. Res. Dev. 2021, 69, 2151–2170. [Google Scholar] [CrossRef]
  12. Ngugi, J.; Goosen, L. Modeling course-design characteristics, self-regulated learning and the mediating effect of knowledge-sharing behavior as drivers of individual innovative behavior. EURASIA J. Math. Sci. Technol. Educ. 2018, 14, em1575. [Google Scholar]
  13. Dang, T. Implementing E-learning in Vietnamese Universities: A Configurational Approach. Ph.D. Dissertation, University of Sheffield, Sheffield, UK, 2014. [Google Scholar]
  14. Bandura, A. Social cognitive theory: An agentic perspective. Annu. Rev. Psychol. 2001, 52, 1–26. [Google Scholar] [CrossRef]
  15. Zamiri, M.; Esmaeili, A. Methods and technologies for supporting knowledge sharing within learning communities: A systematic literature review. Adm. Sci. 2024, 14, 17. [Google Scholar] [CrossRef]
  16. Wasko, M.M.; Faraj, S. Why should I share? Examining social capital and knowledge contribution in electronic networks of practice. MIS Q. 2005, 29, 35–57. [Google Scholar] [CrossRef]
  17. Tsai, H.T.; Bagozzi, R.P. Contribution behavior in virtual communities: Cognitive, emotional, and social influences. Mis Q. 2014, 38, 143–164. [Google Scholar] [CrossRef]
  18. Pang, S.; Bao, P.; Hao, W.; Kim, J.; Gu, W. Knowledge sharing platforms: An empirical study of the factors affecting continued use intention. Sustainability 2020, 12, 2341. [Google Scholar] [CrossRef]
  19. Verona, G.; Prandelli, E.; Sawhney, M. Innovation and virtual environments: Towards virtual knowledge brokers. Organ. Stud. 2006, 27, 765–788. [Google Scholar] [CrossRef]
  20. Gao, D.; Cai, J.; Wu, K. The smart green tide: A bibliometric analysis of AI and renewable energy transition. Energy Rep. 2025, 13, 5290–5304. [Google Scholar] [CrossRef]
  21. Yan, L.; Greiff, S.; Teuber, Z.; Gašević, D. Promises and challenges of generative artificial intelligence for human learning. Nat. Hum. Behav. 2024, 8, 1839–1850. [Google Scholar] [CrossRef]
  22. EIArdeliya, V.; Taylor, J.; Wolfson, J. Exploration of artificial intelligence in creative fields: Generative art, music, and design. Int. J. Cyber IT Serv. Manag. 2024, 4, 40–46. [Google Scholar] [CrossRef]
  23. Bozkurt, A.; Junhong, X.; Lambert, S.; Pazurek, A.; Crompton, H.; Koseoglu, S.; Romero-Hall, E. Speculative futures on ChatGPT and generative artificial intelligence (AI): A collective reflection from the educational landscape. Asian J. Distance Educ. 2023, 18, 53–130. [Google Scholar]
  24. Na-songkhla, J.; Mahakaew, V.; Peytcheva-Forsyth, R. The Emergence of Generative Artificial Intelligence: Enhancing Critical Thinking Skills in ChatGPTIntegrated Cognitive Flexibility Approach. Gener. Artif. Intell. High. Educ. A Handb. Educ. Lead. 2024, 52, 52–63. [Google Scholar]
  25. Cooper, G. Examining science education in ChatGPT: An exploratory study of generative artificial intelligence. J. Sci. Educ. Technol. 2023, 32, 444–452. [Google Scholar] [CrossRef]
  26. Min, H.; Tan, P.X.; Kamioka, E.; Sharif, K.Y. Enhancement of study motivation model by introducing expectancy theory. Int. J. Learn. Teach. 2020, 6, 28–32. [Google Scholar]
  27. Kuhl, J. The expectancy-value approach within the theory of social motivation: Elaborations, extensions, critique. In Expectations and Actions; Routledge: Oxfordshire, UK, 2021; pp. 125–160. [Google Scholar]
  28. Filipovic, J.; Arslanagic-Kalajdzic, M. Social capital theory perspective on the role of academic social networking sites. J. Bus. Res. 2023, 166, 114119. [Google Scholar] [CrossRef]
  29. Parhamnia, F.; Farahian, M.; Rajabi, Y. Knowledge sharing and self-efficacy in an EFL context: The mediating effect of creativity. Glob. Knowl. Mem. Commun. 2022, 71, 293–321. [Google Scholar] [CrossRef]
  30. Lissillour, R.; Cui, Y.; Guesmi, K.; Chen, W.; Chen, Q. Value network and firm performance: The role of knowledge distance and environmental uncertainty. J. Knowl. Manag. 2024, 28, 44–68. [Google Scholar]
  31. Capaldo, A.; Lavie, D.; Messeni Petruzzelli, A. Knowledge maturity and the scientific value of innovations: The roles of knowledge distance and adoption. J. Manag. 2017, 43, 503–533. [Google Scholar]
  32. Gupta, V. An empirical evaluation of a generative artificial intelligence technology adoption model from entrepreneurs’ perspectives. Systems 2024, 12, 103. [Google Scholar] [CrossRef]
  33. Agnihotri, A. Low-cost innovation in emerging markets. J. Strateg. Mark. 2014, 23, 399–411. [Google Scholar] [CrossRef]
  34. Amankwah-Amoah, J.; Abdalla, S.; Mogaji, E.; Elbanna, A.; Dwivedi, Y.K. The impending disruption of creative industries by generative AI: Opportunities, challenges, and research agenda. Int. J. Inf. Manag. 2024, 79, 102759. [Google Scholar] [CrossRef]
  35. Skrbinjek, V.; Vičič Krabonja, M.; Aberšek, B.; Flogie, A. Enhancing teachers’ creativity with an innovative training model and knowledge management. Educ. Sci. 2024, 14, 1381. [Google Scholar] [CrossRef]
  36. De Carolis, B.; Ferilli, S.; Palestra, G. Simulating empathic behavior in a social assistive robot. Multimed. Tools Appl. 2017, 76, 5073–5094. [Google Scholar]
  37. Hsu, M.H.; Chang, C.M.; Yen, C.H. Exploring the antecedents of trust in virtual communities. Behav. Inf. Technol. 2011, 30, 587–601. [Google Scholar] [CrossRef]
  38. Lee, H.Y.; Ahn, H.; Kim, H.K.; Lee, J. Comparative analysis of trust in online communities. Procedia Comput. Sci. 2014, 31, 1140–1149. [Google Scholar] [CrossRef]
  39. Chen, H.L.; Fan, H.L.; Tsai, C.C. The role of community trust and altruism in knowledge sharing: An investigation of a virtual community of teacher professionals. J. Educ. Technol. Soc. 2014, 17, 168–179. [Google Scholar]
  40. Mathibe, I. Expectancy theory and its implications for employee motivation. Acad. Leadersh. Online J. (2003–2012) 2008, 6, 8. [Google Scholar]
  41. Shahzad, M.F.; Xu, S.; Zahid, H. Exploring the impact of generative AI-based technologies on learning performance through self-efficacy, fairness & ethics, creativity, and trust in higher education. Educ. Inf. Technol. 2025, 30, 3691–3716. [Google Scholar]
  42. Erez, A.; Isen, A.M. The influence of positive affect on the components of expectancy motivation. J. Appl. Psychol. 2002, 87, 1055. [Google Scholar] [CrossRef]
  43. Yepes, V.; López, S. The knowledge sharing capability in innovative behavior: A SEM approach from graduate students’ insights. Int. J. Environ. Res. Public Health 2023, 20, 1284. [Google Scholar] [CrossRef]
  44. Chen, C. The effect of leader knowledge hiding on employee voice behavior—The role of leader-member exchange and knowledge distance. Open J. Soc. Sci. 2020, 8, 69–95. [Google Scholar] [CrossRef]
  45. Acar, S.; Burnett, C.; Cabra, J.F. Ingredients of creativity: Originality and more. Creat. Res. J. 2017, 29, 133–144. [Google Scholar] [CrossRef]
  46. Lucas, U.; Tan, P. Assessing levels of reflective thinking: The evaluation of an instrument for use within accounting and business education. In Proceedings of the 1st Pedagogic Research in Higher Education Conference, Liverpool, UK, 2–3 May 2006; pp. 2–3. [Google Scholar]
  47. Kmieciak, R. Trust, knowledge sharing, and innovative work behavior: Empirical evidence from Poland. Eur. J. Innov. Manag. 2021, 24, 1832–1859. [Google Scholar] [CrossRef]
  48. Cheung, C.M.; Lee, M.K. What drives members to continue sharing knowledge in a virtual professional community? The role of knowledge self-efficacy and satisfaction. In International Conference on Knowledge Science, Engineering and Management; Springer: Berlin/Heidelberg, Germany, 2007; pp. 472–484. [Google Scholar]
  49. Yli-Renko, H.; Autio, E.; Sapienza, H.J. Social capital, knowledge acquisition, and knowledge exploitation in young technology-based firms. Strateg. Manag. J. 2001, 22, 587–613. [Google Scholar] [CrossRef]
  50. Ragin, C.C. Redesigning Social Inquiry: Fuzzy Sets and Beyond; University of Chicago Press: Chicago, IL, USA, 2008. [Google Scholar]
  51. Fiss, P.C. Building better causal theories: A fuzzy set approach to typologies in organization research. Acad. Manag. J. 2011, 54, 393–420. [Google Scholar] [CrossRef]
  52. Pappas, I.O.; Woodside, A.G. Fuzzy-set Qualitative Comparative Analysis (fsQCA): Guidelines for research practice in Information Systems and marketing. Int. J. Inf. Manag. 2021, 58, 102310. [Google Scholar] [CrossRef]
  53. Roig-Tiernoa, N.; Gonzalez-Cruz, T.F.; Llopis-Martinez, J. An overview of qualitative comparative analysis: A bibliometric analysis. J. Innov. Knowl. 2016, 2, 15–23. [Google Scholar] [CrossRef]
  54. Freitas, V.; Neto, F. Qualitative comparative analysis (QCA): Usos e aplicações do método. Rev. Política Hoje 2014, 24, 103–117. [Google Scholar]
  55. Rihoux, B.; Ragin, C.C. Configurational Comparative Methods: Qualitative Comparative Analysis (QCA) and Related Techniques; Applied Social Research Methods Series; SAGE: Los Angeles, CA, USA, 2009. [Google Scholar]
  56. Pappas, I.O.; Papavlasopoulou, S.; Mikalef, P.; Giannakos, M.N. Identifying the combinations of motivations and emotions for creating satisfied users in SNSs: An fsQCA approach. Int. J. Inf. Manag. 2020, 53, 102128. [Google Scholar] [CrossRef]
  57. Schneider, C.Q.; Wagemann, C. Standards of good practice in qualitative comparative analysis (QCA) and fuzzy-sets. Comp. Sociol. 2010, 9, 397–418. [Google Scholar] [CrossRef]
  58. Ordanini, A.; Parasuraman, A.; Rubera, G. When the recipe is more important than the ingredients: A qualitative comparative analysis (QCA) of service innovation configurations. J. Serv. Res. 2014, 17, 134–149. [Google Scholar] [CrossRef]
  59. Pappas, I.O.; Mikalef, P.; Giannakos, M.N.; Kourouthanassis, P.E. Explaining user experience in mobile gaming applications: An fsQCA approach. Internet Res. 2019, 29, 293–314. [Google Scholar] [CrossRef]
  60. Hagos, D.H.; Battle, R.; Rawat, D.B. Recent advances in generative ai and large language models: Current status, challenges, and perspectives. IEEE Trans. Artif. Intell. 2024, 5, 5873–5893. [Google Scholar] [CrossRef]
  61. Fong, C.J.; Krause, J.M. Lost confidence and potential: A mixed methods study of underachieving college students’ sources of self-efficacy. Soc. Psychol. Educ. 2014, 17, 249–268. [Google Scholar] [CrossRef]
  62. Schneider, C.Q. Set-Theoretic Multi-Method Research: A Guide to Combining QCA and Case Studies; Cambridge University Press: Cambridge, UK, 2024; Volume 1. [Google Scholar]
  63. Beynon, M.; Pickernell, D.; Battisti, M.; Jones, P. A panel fsQCA investigation on European regional innovation. Technol. Forecast. Soc. Change 2024, 199, 123042. [Google Scholar] [CrossRef]
  64. Li, D.; Zhang, M.; Zhou, S.; Yu, X. Structural equation modeling and validation of virtual learning community constructs based on the Chinese evidence. Interact. Learn. Environ. 2025, 1–16. [Google Scholar] [CrossRef]
  65. Hossain, M.A.; Quaddus, M.; Hossain, M.M.; Gopakumar, G. Data-driven innovation development: An empirical analysis of the antecedents using PLS-SEM and fsQCA. Ann. Oper. Res. 2024, 333, 895–937. [Google Scholar] [CrossRef]
  66. Huang, Y.; Feng, Y.; Gao, D.; Wei, J.; Wu, K. The Power of Knowledge: How Can Educational Competitiveness Improve Urban Energy Efficiency? Sustainability 2025, 17, 6609. [Google Scholar] [CrossRef]
  67. Rogers, D.M.A. Knowledge innovation system: The common language. J. Technol. Stud. 1993, 19, 2–8. [Google Scholar]
  68. Frenz, M.; Ietto-Gillies, G. The impact on innovation performance of different sources of knowledge: Evidence from the UK Community Innovation Survey. Res. Policy 2009, 38, 1125–1135. [Google Scholar] [CrossRef]
Sustainability 17 09060 g001
Figure 1. Configuration effect analysis model of knowledge innovation in the digital learning community.
Figure 1. Configuration effect analysis model of knowledge innovation in the digital learning community.
Sustainability 17 09060 g001
Sustainability 17 09060 g002
Figure 2. XY scatter plot of the overall subordination degree and the knowledge innovation behavior subordination degree.
Figure 2. XY scatter plot of the overall subordination degree and the knowledge innovation behavior subordination degree.
Sustainability 17 09060 g002
Table 1. Constructs, reliability, and validity.
Table 1. Constructs, reliability, and validity.
Construal and MeasurementFactor LoadingReliability and Validity
Creativity
In the ‘Super Star Learn’ community, an AI teaching assistant can bring me new ideas and new experiences.0.839CR = 0.879
AVE = 0.721
Cronbach’s = 0.895
In the ‘Super Star Learn’ community, the ‘AI teaching assistant’ generates innovative content.0.848
Reflexivity
In the ‘Super Star Learn’ community, an AI teaching assistant can not only give answers to my questions but also help me analyze the reasons behind the answers in detail.0.890CR = 0.882
AVE = 0.779
Cronbach’s = 0.898
In the ‘Super Star Learn’ community, the ‘AI teaching assistant’ can self-learn by assessing their behavior to improve their performance the next time it faces similar problems.0.872
Trust relationship (community trust and member trust)
The ‘Super Star Learn’ community management is standardized, and the content is highly reliable.0.850CR = 0.890
AVE = 0.731
Cronbach’s = 0.899
My posts and comments in the ‘Super Star Learn’ community are often liked and positively commented upon by other members.0.842
Incentive mechanism
I can get spiritual rewards such as learning achievement display, an honorary title, or a medal in the ‘Super Star Learn’ community.0.879CR = 0.895
AVE = 0.741
Cronbach’s = 0.919
In the ‘Super Star Learn’ community, I can get material rewards such as learning materials, gifts, or coupons.0.851
Self-efficacy
I believe I can understand the ideas and opinions of other members of the ‘Super Star Learn’ community.0.881CR = 0.859
AVE = 0.757
Cronbach’s = 0.891
I believe in my ability to provide valuable knowledge and ideas to other members of the ‘Super Star Learn’ community.0.857
Expected outcome
I think I can get more respect or make more new friends in the ‘Super Star Learn’ community through my efforts.0.863CR = 0.895
AVE = 0.749
Cronbach’s = 0.909
I think I can get more material rewards in the ‘Super Star Learn’ community through my efforts.0.852
Knowledge distance
In the ‘Super Star Learn’ community, I think I can communicate better with other members on professional issues.0.871CR = 0.923
AVE = 0.741
Cronbach’s = 0.715
In the ‘Super Star Learn’ community, I believe that my knowledge complements that of other members very well.0.882
Knowledge innovation behavior
I often contribute what I have, but I don’t have it in the ‘Super Star Learn’ community.0.863CR = 0.896
AVE = 0.751
Cronbach’s = 0.924
I often put forward new ideas in the process of communicating with other members of the ‘Super Star Learn’ community.0.859
I often follow questions from other members of the ‘Super Star Learn’ community and always come up with newer answers.0.871
Table 2. Descriptive statistics and correlation matrices.
Table 2. Descriptive statistics and correlation matrices.
Mean ValueStandard DeviationCreativityReflexivityTrust RelationshipIncentive MechanismSelf-EfficacyExpected OutcomeKnowledge DistanceKnowledge Innovation Behavior
Creativity5.5101.3990.850
Reflexivity5.4321.4270.765 **0.886
Trust relationship5.4321.4870.331 **0.319 **0.851
Incentive mechanism5.1261.6780.362 **0.347 **0.460 **0.862
Self-efficacy5.1991.6110.359 **0.361 **0.339 **0.585 **0.869
Expected outcome5.1851.5830.391 **0.372 **0.432 **0.608 **0.688 **0.863
Knowledge distance5.3351.1800.691 **0.711 **0.735 **0.487 **0.409 **0.463 **0.856
Knowledge innovation behavior5.1771.6300.372 **0.365 **0.439 **0.631 **0.689 **0.732 **0.501 **0.862
Note: ** p < 0.01; diagonal thickening is the square root of AVE.
Table 3. Necessary condition analysis result.
Table 3. Necessary condition analysis result.
High-Level Knowledge InnovationNon-High-Level Knowledge Innovation
Conditional VariableConsistencyCoverageConsistencyCoverage
Creativity0.8320.7780.6170.716
Non-creativity0.7010.6130.8230.835
Reflexivity0.8340.7620.6640.727
Non reflexivity0.6040.5990.6940.752
Trust relationship0.7930.6950.6920.718
Non-trust relationship0.6350.5730.6960.841
Incentive mechanism0.8790.8110.5380.574
Non-incentive mechanism0.5010.5020.8110.857
Self-efficacy0.8440.7930.4790.599
Non self-efficacy0.5890.5310.7920.847
Expected outcome0.7990.7740.4790.605
Non-expected outcome0.7010.5020.7020.787
Knowledge distance0.8050.7390.5190.603
Non-knowledge distance0.6130.5170.8020.819
Table 4. Configuration analysis of high-level knowledge innovation of learners.
Table 4. Configuration analysis of high-level knowledge innovation of learners.
Individual Cognitive TypeIndividual Cognition–Cultural Environment TypeTechnological Environment–Cultural Environment Type
Conditional ConfigurationConfiguration 1Configuration 2Configuration 3Configuration 4
Creativity ⊗ ⊗ ●
Reflexivity ⊗ ⊗ ⊗ ●
Trust relationship ● ⊗ ●
Incentive mechanism ⊗ ⊗ ●●
Self-efficacy ● ● ⊗
Expected outcome ● ● ● ⊗
Knowledge distance ⊗ ⊗ ●
Consistency0.9830.9710.9860.987
Original coverage0.4510.3030.3970.312
Unique coverage0.2010.0700.1510.134
Consistency of solutions0.961
Coverage of solutions0.652
Note: ● or ● indicates that the condition exists, ⊗ or ⊗ indicates that the condition does not exist; ● or ⊗ indicates a core condition, and ● or ⊗ indicates an edge condition. A blank indicates that the condition may or may not exist.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Huang, Y.; Zhang, Z.; Xu, B.; Zhou, X.; Zhai, J.; Gao, D. Digital Learning Empowering Sustainable Education: Evidence from the Determinants of Chinese College Students’ Knowledge Innovation Capability. Sustainability 2025, 17, 9060. https://doi.org/10.3390/su17209060

AMA Style

Huang Y, Zhang Z, Xu B, Zhou X, Zhai J, Gao D. Digital Learning Empowering Sustainable Education: Evidence from the Determinants of Chinese College Students’ Knowledge Innovation Capability. Sustainability. 2025; 17(20):9060. https://doi.org/10.3390/su17209060

Chicago/Turabian Style

Huang, Yan, Zhihui Zhang, Bingqian Xu, Xinyu Zhou, Jiayu Zhai, and Da Gao. 2025. "Digital Learning Empowering Sustainable Education: Evidence from the Determinants of Chinese College Students’ Knowledge Innovation Capability" Sustainability 17, no. 20: 9060. https://doi.org/10.3390/su17209060

APA Style

Huang, Y., Zhang, Z., Xu, B., Zhou, X., Zhai, J., & Gao, D. (2025). Digital Learning Empowering Sustainable Education: Evidence from the Determinants of Chinese College Students’ Knowledge Innovation Capability. Sustainability, 17(20), 9060. https://doi.org/10.3390/su17209060

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Article metric data becomes available approximately 24 hours after publication online.
Back to TopTop