Abstract
With the continuous advancement of artificial intelligence (AI) technology—particularly the widespread adoption of generative AI—the application of AI tools in the design field has become increasingly prevalent, demonstrating considerable potential for innovation. However, existing research has primarily focused on the acceptance and use of AI technologies by general users, while relatively limited attention has been paid to the behavioral mechanisms of professional creative users, such as digital media designers, in subscribing to advanced AI-assisted drawing tools. Prior studies have often overlooked the nuanced decision-making processes and unique cognitive factors that shape subscription behavior within creative work contexts. To address this gap, the present study draws on the Technology Acceptance Model (TAM), the Unified Theory of Acceptance and Use of Technology (UTAUT), and their extended frameworks, integrating grounded theory’s three-stage coding approach with covariance-based structural equation modeling (CB-SEM) to construct and validate an innovative research framework that bridges qualitative exploration with quantitative validation. Seven key variables—perceived trust, perceived usefulness, perceived ease of use, social influence, output quality, personal innovativeness, and price value—were identified through semi-structured interviews and empirically examined based on data from 394 participants. The results indicate that perceived ease of use indirectly influences subscription intention through perceived usefulness, while output quality affects it through the mediating role of perceived trust. Social influence, price value, and perceived trust all exhibit significant positive effects on subscription intention, whereas personal innovativeness does not demonstrate a statistically significant impact. Theoretically, this study extends the applicability of TAM and UTAUT to the relatively under-researched context of the creative industry, achieves methodological innovation through the integration of qualitative and quantitative approaches, and provides practical guidance and a decision-making foundation for AI tool developers and the intelligent transformation of the creative sector.
Similar content being viewed by others
Introduction
The rapid development of AI is yielding significant economic benefits and causing extensive social impacts across various sectors1. As a seminal technological achievement, AI, fueled by advancements in computer science, marks profound shifts brought about by technological progress2. In the design field, AI’s influence is increasingly apparent, demonstrating substantial potential throughout multiple phases of the design process, such as concept generation, design evaluation, and optimization3. Additionally, digital transformation, a pivotal trend in recent years, is redefining emerging business models, service delivery mechanisms, and revenue structures, thereby significantly boosting enterprise profitability and competitiveness4. Concurrently, digital media art creation and AI technology are advancing in tandem, promoting extensive expansion and interdisciplinary integration within their fields5. The emergence of advanced AI technologies has enabled generative design tools to facilitate new collaborative models in product design, enhancing synergy between human designers and AI and spurring innovation in the development of high-performance and complex products6. For example, Midjourney, a tool that generates artistic images from text inputs, serves various roles, including teacher, mentor, and designer, across disciplines like education, tutoring, assistive processing, and design7. Raina et al.8 highlighted that AI trained to a high level exhibits exceptional capabilities in completing designated design tasks, not only equaling human designers but sometimes surpassing them in efficiency and precision. As AI drawing tools continue to evolve, they have become indispensable in the design industry and will play a crucial role in driving innovation in design and art9.
AI adoption is rapidly expanding across global industries, with its popularity anticipated to rise further in the future10. For instance, the AI drawing software Stable Diffusion quickly attracted over 10 million daily active users following its release11. Similarly, Midjourney reports approximately 1.1 million users as "online and active" at any given time12. Additionally, OpenAI’s ChatGPT has experienced unprecedented growth since its launch in November, exceeding 100 million active users within two months and becoming one of the fastest-growing applications ever13. Compared to their free-tier counterparts, premium versions of AI drawing tools have achieved significant enhancements in functionality and performance, providing increased computing power and advanced intelligence14,15,16. Consequently, a deeper understanding of the interaction mechanisms between users and AI-driven solutions has become imperative for optimizing the design and application of these technologies17. The conversion of free-tier users to paid subscribers and their long-term retention are strategic priorities for ensuring sustainable development and securing competitive advantages in freemium-based services18. Subscription services not only enhance consumer surplus through intensified price competition but also facilitate precise price differentiation across customer groups, optimizing revenue structures and enhancing profitability19. Despite the various forms of servitization, subscription-based business models have become a prominent driver of service growth in traditionally product-centric industries, and serve as a strategic focus in many consumer markets20. Recent statistics show that the global generative AI market is expected to grow from $17.65 billion in 2023 to $25.86 billion in 2024, ultimately reaching an estimated $803.9 billion by 2033, with a compound annual growth rate (CAGR) of 46.5% from 2024 to 203321. Although AI was initially developed as a non-profit project, its shift towards financial sustainability and profitability has become increasingly apparent as the costs associated with large-scale AI applications continue to rise22. In this context, it is particularly important to understand the factors that influence digital media designers’ adoption of premium AI drawing tools to advance the field.
This study rigorously investigates the impact mechanisms behind digital media designers’ adoption of premium AI drawing tools, delineates the primary factors that influence their subscription decisions, and provides theoretical insights to optimize tool design and enhance user satisfaction. By analyzing decision-making pathways and associated impacts, the research offers both theoretical support and practical guidance, enabling developers to devise strategies and encourage innovation in the integration of AI technology within the design domain. For this purpose, 30 digital media designers were recruited via the online platform Discord. Data were gathered through semi-structured interviews, and an in-depth text analysis was performed using a three-tier coding process derived from grounded theory, identifying the key factors shaping designers’ perceptions. Subsequently, a comprehensive theoretical framework was developed by synthesizing the TAM, the UTAUT, and their extended models. This framework was empirically validated using covariance-based CB-SEM. Employing both qualitative and quantitative approaches, the study thoroughly examines the perception factors influencing digital media designers’ engagement with AI drawing tools and the underlying mechanisms affecting their subscription decisions. The findings address significant gaps in existing literature concerning user behavior and payment intentions for AI tools. Furthermore, this research offers substantial theoretical and empirical support for firms aiming to refine promotional strategies for AI design tools, improve user satisfaction, and facilitate broader adoption of AI technology in digital media design.
The structure of this article is as follows: The second section provides the theoretical background, explores the relationship between digital media design and premium AI drawing tools, and details the three-tier coding process based on grounded theory and covariance-based CB-SEM. The third section discusses the research methods, elaborating on the extraction of user perception factors, the construction of research hypotheses and theoretical models, and the detailed implementation of the CB-SEM analysis. The fourth section presents the research findings, including an in-depth discussion of their theoretical and practical significance. It also outlines the research limitations and suggests directions for future developments.
Theoretical background
Digital media design and AI drawing tools
Amidst swift digital advancements, the profound integration of AI and digital media art is transforming perceptions within the fields of art and design23. As a multidisciplinary form, digital media art merges technology and art, building upon theoretical foundations such as information technology and mass communication theory. It possesses significant growth potential, spanning knowledge systems across various domains, including plastic arts, art design, interactive design, computer software, graphics and image processing, and information and communication technology24. Unlike traditional art, digital media art is primarily characterized by the digital display and dissemination of information in the media field25. Previously, designers needed proficiency in tools and techniques to translate their imagination into visual works. Currently, AI drawing tools empower designers to produce high-quality images using mere textual prompts, positioning AI as an essential auxiliary tool in creative processes26. For instance, AI-powered drawing tools such as Disco Diffusion, Midjourney, Stable Diffusion, and DALL-E 2 have attracted considerable attention from designers and artists27. These tools foster creative thinking and visual imagination28, providing innovative support to the design community. They not only broaden the scope of design possibilities but also expedite the creative process, significantly improving design quality and efficiency29.
In recent years, significant advancements in the application of AI to animation and film production have been observed. Reddy et al.30 noted that the introduction of AI technology not only enhances the quality and production efficiency of visual effects but also improves the accuracy and consistency of image processing. Furthermore, it enhances a more immersive audience experience and contributes to the sustainable development of the film and television industry. Likewise, Huang et al.31 highlighted that the application of AI in video production systems optimizes the production process, fulfilling the needs of creators and stimulating their creative enthusiasm and initiative. Moreover, Lee et al.32 investigated the future interactive experiences of integrating AI into daily life, focusing on high-fidelity illustration and human–computer interaction design. User feedback suggests that personalized design and multimodal interaction are essential for improving the user experience. Moreover, the design process must thoroughly consider both user needs and social ethics to ensure the broad acceptance and adaptability of technology. Collectively, these studies illustrate the significant impact of AI technology on the digital media field, underscoring its potential to enhance user experiences and propel industry development.
Grounded theory three-level coding method and CB-SEM
Grounded theory is a qualitative research method designed to develop theories founded on empirical data rather than testing existing theories. The research process utilizes a bottom-up approach, deriving theories from empirical observations through iterative summarization, comparison, correlation, and condensation of collected data, ultimately constructing a theoretical framework33. As an inductive method, grounded theory synthesizes empirical observations with theoretical constructs to derive a comprehensive understanding of the mechanisms underlying complex phenomena from qualitative data. This method significantly increases the ability to explain observed relationships and real-world dynamics34. In the analytical framework of grounded theory, three distinct levels of coding constitute key analytical steps: open coding, axial coding, and selective coding35. Open coding: This process entails analyzing raw data at the level of individual words or phrases, conceptualizing and categorizing information, and enriching these categories through iterative comparisons36. Axial coding: This stage involves clustering and analyzing codes identified during open coding, unveiling potential logical relationships and hierarchical structures. It then organizes these codes into a cohesive, theoretically informed framework37. Selective coding: Expanding on axial coding, this stage examines the interrelationships among categories to develop new theoretical models38. Throughout this analysis, continuous comparison, validation, and reference to relevant literature are crucial to minimize coder bias and ensure that grounded theories authentically articulate the underlying nature of social phenomena39.
SEM is a robust statistical method whose estimation techniques, modeling capabilities, and application areas are rapidly expanding, rendering it an increasingly vital tool in multivariate relationship analysis40. SEM is utilized primarily to explore relationships between latent constructs (factors), commonly characterized by various measurement indicators. It is also known as latent variable analysis or covariance structure analysis and is usually conducted in a confirmatory, rather than exploratory, manner41. Although diverse SEM methods are available, current research predominantly utilizes two approaches: variance-based SEM (PLS-SEM) and covariance-based SEM (CB-SEM)42. CB-SEM is a technique that typically employs software such as AMOS, EQS, LISREL, and MPlus to validate or challenge established hypotheses43. CB-SEM was selected for this study because it effectively validates theoretical hypotheses and manages complex multivariate relationships. Moreover, it provides rigorous statistical testing and model-fitting evaluation44.
In recent years, studies integrating the three-level coding method of grounded theory with CB-SEM have primarily concentrated on areas such as marketing45,46, enterprise management47,48, computer science49,50, education51, and technology acceptance52,53. However, digital media design and user behavior research employing these methodologies remain underexplored. Consequently, this study seeks to address this gap by investigating the crucial factors that influence digital media designers’ adoption of advanced AI drawing tools and elucidating the interplay of these factors in driving subscription behaviors.
Research methods
Extraction of user perceived factors
Data collection
“Discord” was launched in 2015 as a service platform that integrates multimodal chat and voice communication, founded by computer engineers and game developers Jason Citron and Stan Vishnevskiy54. Today, Discord has emerged as a vital platform for global users to communicate, convene, and forge connections55. It has rapidly attracted diverse user groups, encompassing artists, software developers, and study communities, boasting approximately 390 million registered users and 190 million monthly active users56. Two primary reasons for selecting “Discord” as the survey platform include: (1) its widespread use among designers through popular AI drawing tools like Midjourney and Leonardo; (2) its rich ecosystem comprising official AI drawing tool communities, creator groups, and technical exchange forums, which enhances the precision in targeting respondent groups and increases the survey’s validity.
When examining the perceptual factors of digital media designers toward AI drawing tools, it becomes crucial to thoroughly consider the complexity of the user experience and to fully grasp its multidimensional characteristics. Research should not solely depend on secondary sources, such as literature reviews, policy documents, and news reports, but must also collect primary data through comprehensive investigations to ensure the precision and depth of the analysis57. Consequently, this study involved 30 digital media designers with practical experience in utilizing AI drawing tools, recruiting them via advertisements on the Discord social service platform. This study has received formal approval from the Department of Global Convergence at Kangwon National University and has been conducted in strict accordance with relevant guidelines and regulations. All participants in the telephone interviews submitted electronic informed consent forms through an online process. The detailed demographic characteristics of the participants are depicted in Table 1. Semi-structured interviews, valued for their high flexibility, were employed to foster open dialogue around a central theme, minimize interviewer bias, and allow interviewees greater expressive freedom58. The average duration of each interview was 30 min, focusing primarily on user experiences, feedback, and subscription specifics of the AI drawing tools. To ensure data integrity and authenticity, each interview was comprehensively recorded with the consent of the interviewees. The recordings were subsequently transcribed verbatim, culminating in a final interview text of approximately 120,000 words.
Data encoding and analysis
Following the completion of data collection, two-thirds of the textual data were allocated for coding and analysis, while the remaining one-third was reserved for confirming theoretical saturation. The study utilizes a three-tier coding schema based on systematic grounded theory to perform open coding, axial coding, and selective coding of the transcribed text. Assisted by NVivo11 software, a qualitative data analysis tool, the textual data was encoded, and corresponding nodes were established to enhance further analysis.
Initially, in the open coding phase, the raw data were analyzed on a sentence-by-sentence basis, centering around the research topic and purpose to delineate perceptual dimensions tied to specific characteristics. This phase preliminarily identified 42 initial concepts. Then, similar or overlapping concepts were consolidated, recurrent and intersecting ones summarized, and initial concepts with fewer than two occurrences or presenting contradictions were excluded. Consequently, 17 subcategories were developed, reflecting the connotative relationships among concepts, including improvements in design efficiency, support and enhancement of creativity, operational simplicity, fluency of operation, and the peer utilization rate. Representative original statements were allocated to each category, articulating the observed phenomena. Subsequently, in the axial coding phase, the 17 subcategories underwent reclassification, and through a meticulous analysis of their internal relations, 7 higher-order principal categories were delineated. These categories—perceived usefulness, perceived ease of use, social influence, personal innovation, perceived trust, output quality, and price value—epitomize the factors influencing digital media designers’ decisions to adopt the advanced version of AI drawing tools. Ultimately, a theoretical saturation test was applied to one-third of the reserved original interview data. This procedure adhered to the three-tier coding pathway, progressively refining concepts and categories, and systematically comparing and analyzing the outcomes of the subject coding. The findings confirmed that no new concepts, categories, or associations beyond the theoretical framework could be extracted from the residual data, affirming that the primary categories, subcategories, and their associated initial concepts derived from the three-tier coding approach are robust, possessing a coherent logical structure, and indicating that theoretical saturation was attained. The detailed coding results are presented in Table 2.
Building on the preceding research, key determinants influencing the adoption of AI drawing tools from the perspective of digital media designers were pinpointed. These encompass perceived usefulness, perceived ease of use, societal impact, personal innovation, perceived trust, output quality, and price value. Nonetheless, the precise application methods of these factors remain ambiguous. Furthermore, the significant effects of these factors on digital media designers’ subscription to the advanced version of AI drawing tools necessitate additional exploration and analysis.
CB-SEM analysis
Research hypotheses and models
Davis59 proposed the Technology Acceptance Model (TAM) to explain and predict consumers’ willingness and actual behavior in using information technology, emphasizing that perceived usefulness and perceived ease of use are key determinants of technology acceptance. Within the TAM framework, perceived ease of use is conceptualized as a precursor to perceived usefulness59, indicating a significant positive influence on perceived usefulness60. This relationship is grounded in the following rationale: the easier a system is to operate, the more likely it is to enhance individual job performance, thereby reinforcing perceptions of the system’s usefulness61. Empirical studies across various AI application contexts have provided consistent support for this relationship62. For example, research on customer attitudes and behavioral intentions toward robotic restaurant services found that when the interaction process is perceived as easy and convenient, the service is more likely to be regarded as practical and satisfactory, enhancing perceptions of its usefulness63. Similar findings have been reported in studies on AI-integrated customer relationship management systems, where the positive effect of perceived ease of use on perceived usefulness has been empirically validated64. Additionally, research extending the TAM to the domain of mobile food delivery applications has confirmed this relationship60. However, despite substantial empirical support, some studies have not identified a statistically significant relationship between these two constructs65. In light of these inconsistencies, the following hypothesis is proposed:
H1
Perceived ease of use positively influences perceived usefulness.
Within the theoretical framework of the Technology Acceptance Model (TAM), perceived ease of use is defined as the degree to which individuals consider the technology to be effortless to use59,61. As one of the core variables, perceived ease of use has long been regarded as an important factor influencing users’ intentions to adopt and use new technologies62. With the rapid development of AI technologies, the connotation of the Technology Acceptance Model (TAM) has been continuously expanded, and its application scenarios have become increasingly diverse66. Recent studies have indicated that AI can significantly enhance user interaction experiences through intuitive interface design and adaptive functionalities67. Therefore, integrating cutting-edge technologies in generative AI into the TAM framework contributes to a deeper understanding of how perceived ease of use influences digital media designers’ willingness to accept advanced versions of AI drawing tools. In the adoption process of AI drawing tools, perceived ease of use plays a critical role. On the one hand, its intuitive interface design and simplified operational process enable designers to efficiently generate high-quality images; on the other hand, this feature not only enhances design efficiency but also ensures that the output meets professional standards9. In addition, a substantial body of research has further confirmed that perceived ease of use has a significant positive impact on users’ adoption intentions and actual usage behaviors62,68. Based on the above analysis, the following hypothesis is proposed:
H2
Perceived ease of use positively influences purchase usage intention.
Correspondingly, within the theoretical framework of the Technology Acceptance Model (TAM), perceived usefulness refers to the degree to which individuals believe that using the technology can enhance their job performance61. Specifically, when individuals believe that a new tool or technology helps meet their needs and improves their performance, the likelihood of continued use increases significantly, thereby strengthening their intention to adopt the technology69. Multiple studies have consistently shown that perceived usefulness has a positive impact on behavioral intention70,71. For example, a study investigating the intention to use AI chatbots in the hotel and tourism industry found that perceived usefulness significantly increased users’ willingness to adopt the technology72. Similarly, research on AI voice assistants has indicated that perceived usefulness largely influences users’ willingness to use these technologies73. In addition, relevant research on intelligent healthcare services has further demonstrated that perceived usefulness is a key factor in promoting user acceptance of AI technology in healthcare settings74. Therefore, the following hypothesis is proposed:
H3
Perceived usefulness positively influences purchase usage intention.
It is worth noting that Davis59 further proposed that perceived ease of use is an antecedent variable rather than an independent factor parallel to perceived usefulness. In other words, perceived ease of use not only directly influences users’ behavioral intention, but also acts upon perceived usefulness as a prior factor, making it easier for users to recognize the application value of a technology while perceiving its ease of operation75. This causal logic indicates that improving the ease of use of a technology helps indirectly enhance users’ perception of its usefulness, thereby positively affecting their willingness to use it75. Based on this logic, the empirical study conducted by Al-Okaily76 found that perceived usefulness plays a significant mediating role between perceived ease of use and the intention to use AAT. Specifically, perceived ease of use can indirectly enhance users’ adoption intention by increasing perceived usefulness, especially in situations where the technology is perceived as both easy to use and functionally practical, making it more likely to stimulate users’ willingness to use it76. In addition, the study by Zhou et al.77 also supports this viewpoint, further identifying that perceived usefulness serves as a mediator in the relationship between perceived ease of use and behavioral intention. Although this mediating effect is not entirely sufficient, perceived usefulness still plays a key role in explaining how perceived ease of use influences behavioral intention. Therefore, the following hypothesis is proposed:
H3a
Perceived usefulness mediates the relationship between perceived ease of use and purchase usage intention.
Furthermore, Venkatesh and Davis61, based on the Technology Acceptance Model (TAM), proposed the Extended Technology Acceptance Model (TAM2), which emphasizes that social influence and cognitive instrumental processes are the two core variables affecting user perception. Among them, the cognitive instrumental process includes four aspects: job relevance, output quality, result demonstrability, and perceived ease of use. Output quality, as one of the important variables, is defined as the level of positive impact generated by an individual’s use of an information system, namely, the extent to which the outcomes produced by the technology meet the user’s quality expectations and requirements61. This extension of the TAM2 model enables researchers to gain a more comprehensive understanding of the influencing factors in the user technology acceptance process and has therefore been widely applied in empirical research across multiple fields78. In this study, output quality is defined as whether the perceived output of digital media art designers when experiencing AI drawing tools meets their professional standards. If the perceived output quality is higher, the willingness to subscribe to the premium version is more likely to be enhanced. In addition, Chen et al.79, in the study “How Live Streaming Increases Product Sales: The Role of Trust Transfer and Refinement Possibility Model”, pointed out that when consumers perceive high product quality and express satisfaction with the characteristics and performance of the product based on cognitive processes, they are more likely to develop trust in the product on a cognitive basis, thereby increasing their purchase intention. Similarly, Li et al.80 also pointed out that product quality is an important prerequisite for influencing consumer trust building and purchase decision-making. Based on the above theoretical and empirical research, the following hypotheses are proposed:
H4
Output quality positively influences purchase usage intention.
H5
Output quality positively influences perceived trust.
The Unified Theory of Acceptance and Use of Technology (UTAUT) model proposes four core variables that influence user technology adoption: performance expectancy, effort expectancy, social influence, and facilitating conditions68. Among these, social influence refers to the degree to which individuals perceive that important others (such as family and friends) believe they should use the new system, and it is regarded as a key direct determinant of behavioral intention68. Due to the theoretical advantages of this model in explaining user behavioral intentions, UTAUT has been widely applied in studies on information technology adoption behavior81. For example, Wu et al.82 found a significant positive correlation between social influence and college students’ use of AI-assisted learning in the field of AI-assisted education. Similarly, An et al.83 also pointed out that the social influence perceived by English teachers can effectively predict their intention to use AI teaching systems. In addition, the research by Li81 in the field of design further indicates that designers typically enhance their knowledge and skills through communication and discussion with peers, colleagues, and other professionals, and this social interaction process likewise reflects the critical role of social influence in technology adoption. Therefore, the following hypothesis is proposed:
H6
Social influence positively affects purchase usage intention.
Based on the UTAUT theoretical framework, Venkatesh et al.84 further expanded the applicability of UTAUT and proposed the UTAUT2 model, which primarily focuses on technology acceptance and use behavior in consumer contexts. The model adds three key variables—hedonic motivation, price value, and habit—to more comprehensively explain consumers’ technology acceptance and usage patterns. Among these, price value is defined as the cognitive trade-off between consumers’ perceived benefits and the cost of using an application84. In general, individuals tend to pursue the maximization of personal benefits when adopting technology. When they perceive that the overall benefits brought by the technology exceed its usage costs, price value presents a positive effect, thereby significantly enhancing consumers’ intention to use and adoption tendency85. At present, a large number of empirical studies have confirmed the theoretical validity of the UTAUT2 model. It has gradually become a widely used theoretical framework for analyzing technology adoption behavior and emphasizes the significant facilitating role of price value in the technology adoption process, as it can bring users positive experiences and favorable feelings85,86. Therefore, the following hypothesis is proposed:
H7
Price value positively influences purchase usage intention.
In addition, personal innovativeness (PI) is one of the key concepts in the study of information technology innovation adoption, referring to an individual’s inherent tendency toward new technologies. Individuals with higher levels of innovativeness are typically more inclined to be among the first to accept and experiment with emerging technologies87. From another perspective, personal innovativeness can also be understood as the speed and adaptability with which consumers accept new things. Individuals with stronger innovativeness are more likely to quickly and smoothly accept innovative products or emerging technologies, and this trait significantly influences both their willingness and speed of technology adoption60. For example, in the fields of wireless Internet, mobile learning, and mobile payment, studies have confirmed that personal innovativeness has a significant positive impact on user technology adoption88. Meanwhile, other research has also found that consumers’ personal innovativeness positively affects their purchase intention88,89. Based on this, this study defines personal innovativeness as the degree to which digital media designers can quickly and easily accept and adapt to technological innovation when choosing to subscribe to the advanced version of AI drawing tools. Therefore, the following hypothesis is proposed:
H8
Personal innovativeness positively influences purchase usage intention.
Finally, Mayer et al.90 pointed out that trust comprises three core dimensions: ability, benevolence, and integrity. These three factors can enhance consumers’ trust in product providers, thereby reducing uncertainty in the transaction process and promoting the establishment of long-term and stable relationships with them91. Overall, perceived trust can reduce consumers’ risk perception when shopping on platforms, thereby facilitating trust-related behaviors such as purchasing goods and exerting a positive influence on purchase intention92. In the supply–demand relationship, trust has consistently been regarded as one of the key concepts60. Due to the many unique and still unfamiliar characteristics of AI compared with other technologies, its performance and operational mechanisms require further clarification. Therefore, trust plays a crucial role in shaping consumers’ responses to AI companies and their products93. In addition, considering the inherent vulnerability of AI systems, trust is widely regarded as a key prerequisite for their smooth development and successful adoption94. For this study, trust is defined as the perceived level of information authenticity, platform credibility, and the expected trust in future functional performance and service guarantees, as perceived by digital media designers when subscribing to the advanced version of AI drawing tools. In addition, some studies have explored the mediating effect of trust between different quality dimensions and purchase intention95,96. Therefore, trust is not only an important variable influencing users’ final decision-making behavior, but is also regarded as one of the core mediating factors in studies of technology adoption behavior paths, and thus warrants further in-depth investigation95. Based on this, the following hypothesis is proposed:
H9
Perceived trust positively influences purchase usage intention.
H9a
Perceived trust mediates the relationship between output quality and the intention to purchase.
Based on the coding results of qualitative research, this study systematically compared and integrated variables and paths from TAM, UTAUT, and their extended models, identifying eight core variables: perceived usefulness, perceived ease of use, social influence, price value, personal innovation, perceived trust, output quality, and purchase intention. Utilizing these variables and the identified hypothesized paths, this study developed a theoretical model to analyze the relationships among these variables (see Fig. 1).
Research model of this study.
H3a
Perceived ease of use → Perceived usefulness → Purchase usage intention.
H9a
Output quality → Perceived trust → Purchase usage intention.
Questionnaire design and data collection
In this study, the measurement scales for each variable were appropriately adapted from existing validated scales. To enhance the validity and reliability of the scale, two experts in technology acceptance theory and two experts in AI were invited to provide a thorough review and revision. The revised scale eventually comprised 8 dimensions and 27 items (see Appendix A). Experts unanimously concurred that the revised scale exhibits high structural validity and reliability.
This study utilized a five-point Likert scale for questionnaire design, where 1 signified “strongly disagree” and 5 signified "strongly agree." Following the guideline that sample sizes should be 10 to 15 times the number of questionnaire items, the sample size was established. The official test version of the questionnaire included 27 items. Allowing for an estimated dropout rate of about 10% and potential invalid responses, the final projected sample size was between 270 to 450 cases. The distribution and collection of questionnaires occurred via the Chinese professional online research platform “Wenjuanxing” from December 2 to December 10, 2024. All participants were required to have subscribed to and used AI drawing tools actively. This study has received formal approval from the Global Convergence Department of Kangwon National University and has been conducted in strict accordance with relevant guidelines and regulations. Respondents to the online questionnaire were required to carefully read the ethical statement at the top of the questionnaire before proceeding. They could only continue after clicking "Agree." All participants in this stage have provided informed consent. A total of 420 questionnaires were collected. After removing 26 incomplete or untruthful responses, 394 valid questionnaires were retained, resulting in a response validity rate of 93.81%. The basic demographic information and distribution of the respondents are depicted in Table 3.
Given that all data in this study were derived from self-reports by respondents, a potential issue of common method bias (CMB) arises102. To address this, Harman’s single-factor test was utilized to evaluate CMB103. The analysis revealed nine unrotated common factors, each with eigenvalues exceeding 1, which together accounted for 74.224% of the total variance. The predominant factor represented only 9.199% of the variance, remaining below the critical threshold of 40%. As posited by Podsakoff and Organ104, these findings suggest an absence of significant common method bias in this study.
Evaluation of measurement models
Utilizing 394 valid questionnaire responses, this study performed a reliability analysis of eight scale dimensions using SPSS 26.0. The range for Cronbach’s alpha coefficient spans from 0 to 1, with values above 0.7 denoting acceptable reliability105. The results (Table 4) demonstrate that the Cronbach’s alpha coefficients for the dimensions varied from 0.761 to 0.858, all surpassing the 0.7 threshold. This confirms that the measurement items employed in this study exhibit robust internal consistency and satisfy the reliability criteria.
In addition, this study conducted confirmatory factor analysis (CFA) using AMOS 23.0. Although Fornell and Larcker106 recommended that the factor loadings of all measurement items should exceed 0.70, according to the practical standards proposed by relevant studies in recent years52,77,99,107, factor loadings of 0.60 or above can also be considered acceptable for construct validity. In this study, the standardized factor loadings of all measurement items exceeded 0.60, meeting the basic requirements for construct validity. Composite reliability (CR) and average variance extracted (AVE) were used to assess the internal consistency and convergent validity of each construct. As shown in Table 4, the AVE values of all constructs exceeded 0.50 and the CR values exceeded 0.70, indicating strong convergent validity106. Meanwhile, discriminant validity was assessed according to the method proposed by Fornell and Larcker106. As shown in Table 5, the square roots of the AVE values for each construct were greater than the correlation coefficients between that construct and other constructs, further confirming that the measurement model demonstrates good discriminant validity.
Following the standards advocated by Burhan et al.108, Wang and Chen107, and Zhou et al.77, the measurement model demonstrated a satisfactory fit to the data (see Table 6). The fit indices suggest a proficient alignment between the measurement model and the observed data.
Model fitting results and hypothesis testing
The model fit and path significance of the SEM model were assessed using AMOS 23.0. The fit indices presented in Table 6 suggest that the model exhibits a robust fit, characterized by significant validity and appropriateness. The SEM path significance test results (Table 7, Fig. 2) demonstrate that PEU exerts a substantial effect on PU (β = 0.386, p < 0.001), thus supporting H1. Similarly, PEU significantly influences PUI (β = 0.310, p < 0.001), supporting H2. PU has a noteworthy impact on PUI (β = 0.172, p = 0.017), supporting H3. OQ notably affects PUI (β = 0.279, p < 0.001), supporting H4. OQ also significantly affects PT (β = 0.392, p < 0.001), supporting H5. SI significantly influences PUI (β = 0.226, p < 0.001), supporting H6. PV has a significant influence on PUI (β = 0.197, p = 0.003), supporting H7. PI does not exert a significant impact on PUI (β = 0.096, p = 0.097), and thus H8 is not supported. PT significantly affects PUI (β = 0.171, p = 0.014), supporting H9. These findings confirm that PT, PU, PEU, SI, OQ, and PV are critical variables influencing PUI. The standardized path coefficients underscore the relative impact of each variable, ranked as follows: PEU (0.310) > OQ (0.279) > SI (0.226) > PV (0.197) > PU (0.172) > PT (0.171).
Results of model analysis.
To determine whether PU mediates the effect of PEU on PUI and whether PT mediates the effect of OQ on PUI, a bias-corrected percentile bootstrap test was conducted using the Process plugin in SPSS 26.0, comprising 5,000 bootstrap resamples and a 95% confidence interval. A confidence interval excluding 0 suggests a significant mediating effect109. As indicated in Table 8, the bootstrap upper and lower boundaries for both mediating paths exclude 0, substantiating significant mediation for both paths. Therefore, Hypotheses H3a and H9a are supported.
Discussion and inspiration
Discussion
This study aims to explore the principal factors that influence digital media designers’ decisions to subscribe to advanced versions of AI drawing tools. Data were initially gathered through semi-structured interviews, which provided the flexibility necessary for delving into participant insights. The three-level coding method of grounded theory was subsequently employed to identify seven core user perception factors: PT, PU, PEU, SI, OQ, PI, and PV. A theoretical model was then constructed, and research hypotheses were formulated by synthesizing variables and pathways discerned in empirical studies, drawing on the TAM, UTAUT, and their extended models. Finally, empirical tests were performed using CB-SEM to confirm the relationships and path effects among variables, thereby elucidating the impact mechanism of these factors on user subscription intentions.
The results of the SEM analysis indicate that PEU significantly influences both PU and PUI positively (H1, H2), and PU also positively impacts PUI significantly (H3). Additionally, OQ significantly enhances both PUI and PT’s effectiveness (H4, H5), and PU mediates the relationship between PEU and PUI (H3a). This finding is consistent with the fundamental principles of the TAM and its extended version, TAM2. Digital media designers prioritize factors that align closely with their professional requirements. These factors are critical in their decision-making process when subscribing to the advanced version of AI drawing tools. Enhancing design efficiency is a primary motivator for managing high-intensity, short-cycle projects. AI tools swiftly generate varied graphic schemes and visual effects, enabling designers to concentrate on essential creative tasks. Creative support and enhancement are vital necessities in digital media design. Through generative design, AI tools provide innovative and pioneering creative solutions, facilitating unique visual expressions. Usability and operational fluidity are crucial. Digital media design necessitates the integration of tools from different fields, such as video editing, UI design, and 3D modeling. Streamlining operations reduces learning costs and significantly improves workflow efficiency. Moreover, the precision of output directly influences the quality of work across various media platforms, including the visual impact of social media advertisements, the smoothness of user experience in interactive interfaces, and the presentation of details in dynamic animations. AI tools must satisfy designers’ professional needs for high resolution and detail precision to adjust to complex design scenarios. Another essential criterion for evaluating tool value is the applicability of design results. Digital media work must seamlessly adapt to various scenarios and devices, including mobile, desktop, and large-screen setups. This adaptability directly impacts the final output’s effectiveness. Furthermore, the diversity of creative outputs enables designers to sustain a distinctive visual style in competitive markets. AI tools support this by generating multiple styles and themes, assisting designers in continuously creating novel visual languages. In summary, these factors collectively determine the principal considerations for digital media designers when selecting AI drawing tools, highlighting the industry’s significant demand for precision, compatibility, and creative support in AI-powered tools for digital media design.
SI and PV significantly positively impact PUI (H6, H7), consistent with the foundational assumptions of the UTAUT and its advanced version, UTAUT2. This implies that in deciding to subscribe to an advanced AI drawing tool, digital media designers meticulously evaluate crucial factors including the widespread use among peers, industry recognition, awareness of trends, and cost-effectiveness. In the collaborative environment of the digital media sector, peer use’s popularity is crucial. Widely adopted tools ensure seamless integration across teams and project workflows, aiding designers in quickly becoming proficient with the tools and enhancing efficiency through the exchange of experiences with peers. Consistency within the team is especially vital for progressing projects in areas such as advertising creativity, game design, and multimedia production. Industry recognition lays a trust foundation for tool selection, with endorsements from authoritative figures or leading designers acting as principal references. Such acknowledgments confirm the tool’s efficacy in high-caliber digital media endeavors, for instance, in film and television special effects production or interactive media development. Trend awareness influences designers’ competitive position in the market. With rapid technological progress in the digital media industry, designers are compelled to adopt cutting-edge AI tools to sustain their competitiveness in areas like dynamic posters, virtual reality design, and immersive content creation. Budget-conscious digital media designers prioritize cost-effectiveness and pricing transparency. AI tools that offer a rich array of features at reasonable and transparent prices facilitate optimal investment in multi-platform collaborations, such as for budget management in multi-screen adaptation design or cross-platform content publishing.
PT significantly positively influences PUI (H9) and mediates the relationship between OQ and PUI (H9a), in alignment with the results reported by Zhu et al.92, Qalati et al.95, and Zhao et al.96. When deciding to subscribe to the advanced version of AI drawing tools, digital media designers prioritize privacy and data security, timely customer service responses, and the trustworthiness of the platform. These concerns are closely associated with the professional attributes of the digital media industry. Digital media designers often manage core content related to brand visual identity, advertising creativity, interactive media development, and unreleased products, which are crucial to clients’ marketing strategies. Data breaches may result in notable damage to clients’ brand image and designers’ professional reputations. Therefore, AI tools must ensure robust encryption mechanisms and privacy protection measures while maintaining strict control over sensitive data access. Furthermore, multi-departmental collaboration and intricate production cycles are prevalent in digital media design projects. When facing tight delivery deadlines, technical issues can directly impede project progress. Real-time online support and responsive customer service can substantially minimize delays and promote efficient task completion. Moreover, platform trustworthiness holds particular significance in the digital media industry. Designers tend to trust platforms that are extensively utilized in high-level projects and possess strong reputations. Such platforms must continuously refine their features and provide clear copyright ownership policies for AI-generated works, which are essential for successful project delivery and maintaining customer relationships. A platform with a high reputation, characterized by professionalism and accountability, fosters a secure and effective creative environment, empowering designers to concentrate on brand storytelling, user experience, and innovative communication strategies.
The findings of this study did not substantiate a positive effect of PI on PUI (H8), which contrasts with prior research88,89,110 indicating that PI is a crucial element influencing users’ purchase usage intentions. Nevertheless, the results of this study reveal that PI does not significantly affect digital media designers’ intention to subscribe to advanced versions of AI drawing tools. This absence of significant impact may be closely related to the unique professional characteristics of the industry. In a professional setting where customer needs are paramount and high-quality creative solutions must be delivered within strict time and budgetary constraints, designers are more inclined to depend on industry trends, peer recommendations, and practical user experiences rather than personal exploration. Specifically, aspects such as technical complexity, learning costs, and compatibility with existing design software and workflows are pivotal considerations for AI drawing tools. Seamless integration of these tools with mainstream design platforms crucially influences their efficacy within the design workflow. Moreover, AI tools must efficiently generate content that symbolizes brand positioning, adapt to multi-platform outputs (e.g., social media, video advertising, interactive applications), and enable robust adjustment and iteration capabilities. These aspects typically take precedence over the tools’ innovative features. Furthermore, in the digital media sector, industry recognition and peer endorsements often constitute primary criteria for designers when selecting tools, as these elements validate the tools’ reliability and practicality. Designers generally prefer tools that are widely accepted and validated for collaborative projects with clients or teams, aiming to ensure workflow efficiency and the stability of deliverables. Thus, the influence of individual innovation among this professional group is potentially subordinated by the more pressing factors, underscoring digital media designers’ substantial need for practical applicability, efficiency, and compatibility in AI tools.
Theoretical significance
The theoretical contributions of this study are reflected in the following aspects:
First, this study introduces a novel framework for systematically analyzing user perceptions and behaviors by integrating grounded theory’s three-level coding method with covariance-based structural equation modeling (CB-SEM). Unlike prior research that relied exclusively on the classical Technology Acceptance Model (TAM) or the Unified Theory of Acceptance and Use of Technology (UTAUT), this study investigates context-specific and critical user perception factors relevant to digital media designers—namely, perceived trust, output quality, and price value—identified through the grounded theory coding process. These variables have not yet been fully incorporated into existing technology acceptance frameworks. Thus, this study deepens and extends classical technology acceptance theories by enriching the conceptual foundation of TAM and UTAUT within the context of digital media design and AI-assisted creative applications.
Secondly, this study clarifies the mediating relationships among key variables in the classical technology acceptance model, specifically verifying the mediating role of perceived usefulness between perceived ease of use and subscription intention. This finding not only deepens the theoretical understanding of the interplay among core constructs within the technology adoption mechanism, but also provides empirical support for the theoretical extension of TAM in emerging technology domains, particularly in the context of generative AI tools. Furthermore, this study confirms the significant mediating role of perceived trust between output quality and users’ behavioral intentions, thereby enriching the theoretical discourse on trust mechanisms within existing technology acceptance theories and revealing the distinct theoretical logic underlying AI tool acceptance contexts.
Furthermore, this study adopts an innovative integration of grounded theory’s three-level coding method and covariance-based structural equation modeling (CB-SEM), highlighting the unique advantages of a hybrid methodological approach in theory development. On one hand, grounded theory’s three-level coding provides a rigorous methodological foundation for deriving theory from empirical data, effectively minimizing researcher subjectivity and enhancing the practical relevance of the resulting constructs. On the other hand, the CB-SEM technique offers robust statistical validation for the newly developed theoretical model, ensuring the reliability and rigor of the proposed theoretical pathways. This integrated approach not only represents a methodological innovation, but also exemplifies a comprehensive trajectory of theory development—from data-driven construction to empirical validation—offering valuable paradigmatic guidance for future theoretical research.
Finally, this study focuses on the specific professional group of digital media designers, uncovering their distinct cognitive and decision-making mechanisms in the context of subscribing to advanced versions of AI-assisted drawing tools. Unlike general consumers, this group places greater emphasis on the practical functions of such tools—particularly in enhancing design efficiency, supporting creative processes, and ensuring output quality—within high-intensity, short-cycle, and creativity-driven work environments. The theoretical model developed in this study incorporates key variables such as perceived usefulness, perceived ease of use, output quality, price value, perceived trust, and social influence, thereby systematically identifying the primary drivers of subscription intention among digital media designers engaged in complex design tasks. This expands the application scope of technology acceptance theories within the creative professions. Furthermore, by conducting an in-depth analysis of functional fit and usage barriers across different stages of the design process, this study further enriches the theoretical framework of user behavior research in digital media design.
Practical significance
This study offers multiple practical implications:
First, it systematically identifies, for the first time, the key factors influencing digital media designers’ intentions to subscribe to advanced versions of AI-assisted drawing tools, providing targeted empirical evidence for product development and optimization. The findings indicate that perceived ease of use and output quality are core considerations in designers’ decision-making processes. This suggests that developers should prioritize enhancing operational intuitiveness and the professional quality of image generation when iterating product features. Specifically, functionalities such as resolution switching, one-click format export, and preset style templates can be implemented to meet the professional demands of diverse application scenarios, including social media short videos, advertising posters, and UI prototypes. Simultaneously, the user interface layout and prompt system should be optimized to improve tool learnability, and the diversity and refinement of image styles should be enhanced to meet the high-quality standards of creative work. These optimizations can effectively support advertising designers, UI/UX designers, and social media content creators in improving work efficiency and producing more expressive and professional outcomes.
Secondly, this study provides empirical evidence to support the development of targeted marketing and business strategies by AI tool companies. The results confirm that social influence and price value significantly affect users’ subscription intentions, suggesting that companies can design strategic initiatives tailored to the communication dynamics of the designer community. For example, influential digital media designers can be invited to participate in product testing and promotional campaigns, showcasing real-world use cases and creative outcomes through platforms such as online designer communities or Discord. These strategies can help build brand reputation, enhance platform visibility, and foster trust among potential users. In addition, based on the importance of price value, companies should consider offering more flexible subscription models to meet the diverse needs of different user segments. Options may include project-based billing, short-term trial access, or tiered programs such as student, professional, and enterprise plans. These models can better accommodate the varying budget constraints and usage expectations of freelance designers, small studios, and large organizations, thereby enhancing adoption and long-term user engagement.
Furthermore, this study underscores the pivotal role of perceived trust in subscription decision-making, offering practical guidance for enterprises aiming to build trustworthy AI-based design platforms. Digital media designers frequently manage clients’ brand proposals, promotional materials, and visual content for confidential projects, making their work highly sensitive. Consequently, they are particularly attentive to issues such as data security, copyright ownership, and platform stability. To enhance user trust, enterprises should communicate privacy protection policies and security standards clearly and visibly on the platform interface. In addition, implementing a responsive and intelligent customer support system—such as embedded AI-driven Q&A modules and access to dedicated technical consultants during working hours—can help users obtain timely assistance during the creative process and improve the overall user experience. By strengthening data encryption, optimizing support infrastructure, and cultivating a professional and reliable platform image, companies can effectively reinforce designers’ trust in the platform, thereby increasing both their subscription intentions and long-term usage engagement.
Finally, this research holds significant practical implications for advancing the intelligent transformation and human–machine collaboration within the digital media industry. With the widespread adoption of AI-generated images in brand visuals, illustration drafts, short-form video assets, and other creative contexts, AI tools have increasingly become essential creative “assistants” for digital media designers. The model proposed in this study can serve as a valuable reference for design institutions, educational organizations, and tool developers to optimize training programs and collaborative workflows. It can also support designers in more effectively integrating AI tools into professional creative practices, thereby accelerating the digital and intelligent evolution of the industry as a whole.
Limitations and future research directions
This study has several limitations that require further exploration. First, it adopted a combination of semi-structured interviews and questionnaire surveys, and the data obtained primarily relied on participants’ self-reports, which may be influenced by over- or under-reporting of their behaviors or perceptions. It is suggested that future research adopt a longitudinal design to track the dynamic changes in user behavior and perceptions over time, enabling an in-depth analysis of the temporal evolution of subscription behavior and long-term influencing factors. In addition, although this study integrated key variables from the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT) to construct a relatively systematic analytical model, there are still limitations in the value perception dimension. At present, only the utilitarian factor of “price value” is included, while other value types that may influence subscription intention—such as social value and emotional value—are not covered. Future research could further expand interview scope and sample sources to explore the comprehensive impact of multiple value dimensions on user behavior, thereby enhancing the theoretical extension and explanatory power of the model. Meanwhile, regarding the antecedent variables of trust, only “output quality” was included in this study. Future research could further investigate whether variables such as “usefulness” and “innovativeness” indirectly influence trust through mediating pathways, to improve the theoretical model of trust formation mechanisms. Finally, in response to the finding that personal innovativeness did not have a significant effect on purchase intention, future studies could further examine the boundary conditions and underlying mechanisms that affect the role of innovativeness, and reveal how this variable functions in different contextual settings.
Data availability
The datasets generated during or analysed during the current study are available in the Figshare repository, https://doi.org/10.6084/m9.figshare.28190297.
References
Li, B. et al. Trustworthy AI: From principles to practices. ACM Comput. Surv. 55(9), 1–46 (2023).
Gong, Y. Application of virtual reality teaching method and artificial intelligence technology in digital media art creation. Ecol. Inform. 63, 101304 (2021).
Williams, G., Meisel, N. A., Simpson, T. W. & McComb, C. Design repository effectiveness for 3D convolutional neural networks: Application to additive manufacturing. J. Mech. Design 141(11), 111701 (2019).
World Economic Forum (WEF). World Economic Forum Annual Meeting. https://www.weforum.org/agenda/2023/01/davos23-digital-transformation-investment-priority-downturn/ (2023).
Qian, J. Research on artificial intelligence technology of virtual reality teaching method in digital media art creation. J. Internet Technol. 23(1), 125–132 (2022).
Saadi, J. I. & Yang, M. C. Generative design: Reframing the role of the designer in early-stage design process. J. Mech. Design 145(4), 041411 (2023).
Chiu, T. K. The impact of Generative AI (GenAI) on practices, policies and research direction in education: A case of ChatGPT and Midjourney. Interact. Learn. Envir. 32(10), 6187–6203 (2023).
Raina, A., McComb, C. & Cagan, J. Learning to design from humans: Imitating human designers through deep learning. J. Mech. Design 141(11), 111102 (2019).
Fan, P. & Jiang, Q. Exploring the factors influencing continuance intention to use AI drawing tools: Insights from designers. Systems 12(3), 68 (2024).
Lee, Y. S., Kim, T., Choi, S. & Kim, W. When does AI pay off? AI-adoption intensity, complementary investments, and R&D strategy. Technovation 118, 102590 (2022).
Kemper, J. Stable Diffusion startup Stability AI needs new capital. Available at: https://the-decoder.com/stable-diffusion-startup-stability-ai-needs-new-capital/ (2023).
Wilson, A. Midjourney statistics: Users, polls, & growth [Oct 2023]. Approachable AI. Retrieved from https://approachableai.com/midjourney-statistics/ (2023).
Harrison Dupré, M. ChatGPT’s Explosive Popularity Makes It the Fastest-Growing App in Human History. Futurism https://futurism.com/the-byte/chatgpts-fastest-growing-app-human-history (2023).
Yu, H. Reflection on whether Chat GPT should be banned by academia from the perspective of education and teaching. Front. Psychol. 14, 1181712 (2023).
Sahu, P. K., Benjamin, L. A., Singh Aswal, G. & Williams-Persad, A. ChatGPT in research and health professions education: Challenges, opportunities, and future directions. Postgrad. Med. J. 100(1179), 50–55 (2024).
Frenkel, M. E. & Emara, H. ChatGPT-35 and-40 and mechanical engineering: Examining performance on the FE mechanical engineering and undergraduate exams. Comput. Appl. Eng. Educ. 32(6), e22781 (2024).
Jo, H. Uncovering the reasons behind willingness to pay for ChatGPT-4 premium. Int. J. Hum. Comput. Interact. 41(2), 994–1009 (2024).
Mäntymäki, M., Islam, A. N. & Benbasat, I. What drives subscribing to premium in freemium services? A consumer value-based view of differences between upgrading to and staying with premium. Inform. Syst. J. 30(2), 295–333 (2020).
Wang, R., Dada, M. & Sahin, O. Pricing ancillary service subscriptions. Manage. Sci. 65(10), 4712–4732 (2019).
Kowalkowski, C. & Ulaga, W. Subscription offers in business-to-business markets: Conceptualization, taxonomy, and framework for growth. Ind. Market. Manag. 117, 440–456 (2024).
Precedence Research. Generative AI Market (By Component: Software, Services; By Technology: Generative Adversarial Networks (GANs), Transformers, Variational Auto-encoders, Diffusion Networks; By End-Use: Automotive & Transportation, BFSI, Media & Entertainment, IT & Telecommunication, Healthcare, Others) - Global Industry Analysis, Size, Share, Growth, Trends, Regional Outlook, and Forecast 2024–2033. https://www.precedenceresearch.com/generative-ai-market (2024).
Meyer, J. G. et al. ChatGPT and large language models in academia: Opportunities and challenges. Biodata Min. 16(1), 20 (2023).
Chen, C. Study on the innovative development of digital media art in the context of artificial intelligence. Comput. Intel. Neurosc. https://doi.org/10.1155/2022/1004204 (2022).
Yu, W. & Jiang, T. Research on the direction of innovation and entrepreneurship education reform within the digital media art design major in the digital economy. Front. Psychol. 12, 719754 (2021).
Tian, H. Application and analysis of artificial intelligence graphic element algorithm in digital media art design. Mob. Inf. Syst. https://doi.org/10.1155/2022/6946616 (2022).
Lyu, Y., Wang, X., Lin, R. & Wu, J. Communication in human–AI co-creation: Perceptual analysis of paintings generated by text-to-image system. Appl. Sci. 12(22), 11312 (2022).
Zhang, Y. & Liu, C. Unlocking the potential of artificial intelligence in fashion design and E-commerce applications: The case of midjourney. J. Theor. Appl. El. Comm. 19(1), 654–670 (2024).
Eapen, T., Finkenstadt, D. J., Folk, J. & Venkataswamy, L. How generative AI can augment human creativity. Harvard Bus. Rev. 101(4), 56–64 (2023).
Ma, H. & Li, N. Exploring user behavioral intentions and their relationship with AI design tools: A future outlook on intelligent design. IEEE Access 12, 149192–149205 (2024).
Reddy, V. S., Kathiravan, M. & Reddy, V. L. Revolutionizing animation: Unleashing the power of artificial intelligence for cutting-edge visual effects in films. Soft Comput. 28(1), 749–763 (2024).
Huang, Y. et al. Recent advances in artificial intelligence for video production system. Enterp. Inf. Syst-Uk. 17(11), 2246188 (2023).
Lee, S., Lee, M. & Lee, S. What if artificial intelligence become completely ambient in our daily lives? Exploring future human-ai interaction through high fidelity illustrations. Int. J. Hum. Comput. Interact. 39(7), 1371–1389 (2023).
Glaser, B. G. & Strauss, A. L. The discovery of grounded theory: Strategies for qualitative research. 271 (1967).
Wuelser, G. & Pohl, C. How researchers frame scientific contributions to sustainable development: A typology based on grounded theory. Sustain. Sci. 11, 789–800 (2016).
Strauss, A. & Corbin, J. Basics of qualitative research techniques 2nd edn, 1–312 (Sage, 1998).
Zhang, J. X. et al. Influencing factors of urban innovation and development: A grounded theory analysis. Environ. Dev. Sustain. 25(3), 2079–2104 (2023).
Strauss, A. & Corbin, J. Basics of qualitative research Vol. 15, 61–110 (Sage, 1990).
Corbin, J. & Strauss, A. Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory Vol. 3, 240–359 (Sage Publications Inc, 2008).
Ma, K., Gao, M., Guida, F. E. & He, R. Understanding the influencing factors and mechanism of social compensation for Chinese older adults using social media in the context of smart home: A qualitative analysis. Front. Public Health 11, 1174920 (2023).
Lei, P. W. & Wu, Q. Introduction to structural equation modeling: Issues and practical considerations. Educ. Meas-Issues. Pra. 26(3), 33–43 (2007).
Dash, G. & Paul, J. CB-SEM vs PLS-SEM methods for research in social sciences and technology forecasting. Technol. Forecast. Soc. 173, 121092 (2021).
Henseler, J. & Schuberth, F. Should PLS become factor-based or should CB-SEM become composite-based? Both!. Eur. J. Inform. Syst. https://doi.org/10.1080/0960085X.2024.2357123 (2024).
Ivanov, S., Soliman, M., Tuomi, A., Alkathiri, N. A. & Al-Alawi, A. N. Drivers of generative AI adoption in higher education through the lens of the Theory of Planned Behaviour. Technol. Soc. 77, 102521 (2024).
Vinkóczi, T., Rácz, É. H. & Koltai, J. P. Exploratory analysis of zero waste theory to examine consumer perceptions of sustainability: A covariance-based structural equation modeling (CB-SEM). Clean. Waste Syst. 8, 100146 (2024).
Zhang, M., Li, Y., Sun, L. & Moustapha, F. A. Integrated store service quality measurement scale in omni-channel retailing. Int. J. Retail Distrib. 50(7), 839–859 (2022).
Oh, T. T. & Pham, M. T. A liberating-engagement theory of consumer fun. J. Consum. Res. 49(1), 46–73 (2022).
Zhao, X., Peng, B., Zheng, C. & Wan, A. Business model innovation risk factors based on grounded theory: A multiple-case analysis of cold chain logistics companies in China. Manag. Decis. Econ. 43(6), 2108–2118 (2022).
Khan, S. A. R., Sheikh, A. A. & Tahir, M. S. Corporate social responsibility—An antidote for sustainable business performance: Interconnecting role of digital technologies, employee eco-behavior, and tax avoidance. Environ. Sci. Pollut. R. 31(3), 4365–4383 (2024).
Fang, S., Han, X. & Chen, S. The impact of tourist–robot interaction on tourist engagement in the hospitality industry: A mixed-method study. Cornell Hosp. Q. 64(2), 246–266 (2023).
Li, J., Zhang, S. & Ao, W. Why is instant messaging not instant? Understanding users’ negative use behavior of instant messaging software. Comput. Hum. Behav. 142, 107655 (2023).
Zhu, J. et al. Academic adaptation of international students in China: Evidence from the grounded theory and structure equation model. Sustainability 15(1), 692 (2022).
Samaddar, K. & Mondal, S. AR and VR-based travel: A responsible practice towards sustainable tourism. Int. J. Tour. Cities. 10(1), 105–128 (2024).
Kumar, A. & Shankar, A. The bold decision to go “all in”: Understanding the reasons behind consumers’ willingness to migrate to the metaverse. Psychol. Market. 41(8), 1769–1791 (2024).
Robinson, B. Governance on, with, behind, and beyond the Discord platform: A study of platform practices in an informal learning context. Learn. Media Technol. 48(1), 81–94 (2023).
Our Story. (n.d.). Discord. https://discord.com/company (2024).
The Small Business Blog. How many people use Discord? (Discord Statistics 2024). https://thesmallbusinessblog.net/discord-statistics/ (2024).
Chen, C., Yang, J., Gao, J. & Chen, W. An observation over the rural-urban re-connecting process based on the Alternative Food Network (AFN) in China——From the perspective of ‘social capital’. Habitat Int. 130, 102708 (2022).
McIntosh, M. J. & Morse, J. M. Situating and constructing diversity in semi-structured interviews. Glob. Qualit. Nurs. Res. 2, 2333393615597674 (2015).
Davis, F. D. Perceived usefulness, perceived ease of use, and user acceptance of information technology. Mis. Quart. 13(3), 319–340 (1989).
An, S., Eck, T. & Yim, H. Understanding consumers’ acceptance intention to use mobile food delivery applications through an extended technology acceptance model. Sustainability 15(1), 832 (2023).
Venkatesh, V. & Davis, F. D. A theoretical extension of the technology acceptance model: Four longitudinal field studies. Manage. Sci. 46(2), 186–204 (2000).
Ma, J. et al. Exploring user adoption of ChatGPT: A technology acceptance model perspective. Int. J. Hum. Comput. Interact. 41(2), 1431–1445 (2025).
Choe, J. Y., Kim, J. J. & Hwang, J. Innovative robotic restaurants in Korea: Merging a technology acceptance model and theory of planned behaviour. Asian J. Technol. Inno. 30(2), 466–489 (2022).
Chatterjee, S., Chaudhuri, R., Vrontis, D., Thrassou, A. & Ghosh, S. K. Adoption of artificial intelligence-integrated CRM systems in agile organizations in India. Technol. Forecast Soc. 168, 120783 (2021).
Chen, C. C. & Tsai, J. L. Determinants of behavioral intention to use the personalized location-based mobile tourism application: An empirical study by integrating TAM with ISSM. Fut. Gener. Comp Sy. 96, 628–638 (2019).
Noor, N., Hill, S. R. & Troshani, I. Developing a service quality scale for artificial intelligence service agents. Eur. J. Market. 56(5), 1301–1336 (2022).
Topsakal, Y. How familiarity, ease of use, usefulness, and trust influence the acceptance of generative artificial intelligence (AI)-assisted travel planning. Int. J. Hum. Comput. Interact. https://doi.org/10.1080/10447318.2024.2426044 (2024).
Venkatesh, V., Morris, M. G., Davis, G. B. & Davis, F. D. User acceptance of information technology: Toward a unified view. Mis. Quart. 27(3), 425–478 (2003).
Cambra-Fierro, J. J., Blasco, M. F., López-Pérez, M. E. E. & Trifu, A. ChatGPT adoption and its influence on faculty well-being: An empirical research in higher education. Educ. Inf. Technol. 30, 1517–1538 (2024).
Agarwal, R. & Karahanna, E. Time flies when you’re having fun: Cognitive absorption and beliefs about information technology usage. Mis. Quart. 24(4), 665–694 (2000).
Oyman, M., Bal, D. & Ozer, S. Extending the technology acceptance model to explain how perceived augmented reality affects consumers’ perceptions. Comput. Hum. Behav. 128, 107127 (2022).
Pillai, R. & Sivathanu, B. Adoption of AI-based chatbots for hospitality and tourism. Int. J. Contemp. Hosp. M. 32(10), 3199–3226 (2020).
Cai, R., Cain, L. N. & Jeon, H. Customers’ perceptions of hotel AI-enabled voice assistants: Does brand matter?. Int. J. Contemp. Hosp. M. 34(8), 2807–2831 (2022).
Liu, K. & Tao, D. The roles of trust, personalization, loss of privacy, and anthropomorphism in public acceptance of smart healthcare services. Comput. Hum. Behav. 127, 107026 (2022).
Wang, S. et al. The influence of AR on purchase intentions of cultural heritage products: The TAM and flow-based study. Appl. Sci. 14(16), 7169 (2024).
Al-Okaily, M. Attitudes toward the adoption of accounting analytics technology in the digital transformation landscape. J. Account Organ. Chan. https://doi.org/10.1108/JAOC-04-2024-0127 (2024).
Zhou, C., Liu, X., Yu, C., Tao, Y. & Shao, Y. Trust in AI-augmented design: Applying structural equation modeling to AI-augmented design acceptance. Heliyon 10(1), 23305 (2024).
Pagnanelli, M. A., Colamatteo, A., Cassia, F. & Sansone, M. Unveiling user intentions: An integrated analysis of hedonic motivation and service quality in e-grocery technology acceptance. Int. J. Qual. Reliab. Ma. https://doi.org/10.1108/ijqrm-02-2024-0062 (2025).
Chen, C. D., Zhao, Q. & Wang, J. L. How livestreaming increases product sales: Role of trust transfer and elaboration likelihood model. Behav. Inform. Technol. 41(3), 558–573 (2022).
Li, X., Hess, T. J. & Valacich, J. S. Using attitude and social influence to develop an extended trust model for information systems. ACM SIGMIS Datab. DATAB Adv. Inform. Syst. 37(2–3), 108–124 (2006).
Li, W. A study on factors influencing designers’ behavioral intention in using AI-generated content for assisted design: Perceived anxiety, perceived risk, and UTAUT. Int. J. Hum. Comput. Interact. 41(2), 1064–1077 (2025).
Wu, W., Zhang, B., Li, S. & Liu, H. Exploring factors of the willingness to accept AI-assisted learning environments: An empirical investigation based on the UTAUT model and perceived risk theory. Front. Psychol. 13, 870777 (2022).
An, X. et al. Modeling English teachers’ behavioral intention to use artificial intelligence in middle schools. Educ. Inf. Technol. 28(5), 5187–5208 (2023).
Venkatesh, V., Thong, J. Y. & Xu, X. Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. Mis Quart. 36(1), 157–178 (2012).
de Blanes, G., Sebastián, M., Sarmiento Guede, J. R. & Antonovica, A. Application and extension of the UTAUT2 model for determining behavioral intention factors in use of the artificial intelligence virtual assistants. Front Psychol. 13, 993935 (2022).
Zhu, W. et al. Could AI ethical anxiety, perceived ethical risks and ethical awareness about AI influence university students’ use of generative AI products? An ethical perspective. Int. J. Hum. Comput. Interact. 41(1), 742–764 (2025).
Agarwal, R. & Prasad, J. A conceptual and operational definition of personal innovativeness in the domain of information technology. Inform. Syst. Res. 9(2), 204–215 (1998).
He, X., Zhan, W. & Hu, Y. Consumer purchase intention of electric vehicles in China: The roles of perception and personality. J. Clean. Prod. 204, 1060–1069 (2018).
Chauhan, H., Pandey, A., Mishra, S. & Rai, S. K. Modeling the predictors of consumers’ online purchase intention of green products: The role of personal innovativeness and environmental drive. Environ. Dev. Sustain. 23, 16769–16785 (2021).
Mayer, R. C., Davis, J. H. & Schoorman, F. D. An integrative model of organizational trust. Acad. Manage. Rev. 20(3), 709–734 (1995).
Suh, B. & Han, I. The impact of customer trust and perception of security control on the acceptance of electronic commerce. Int. J. Electron. Comm. 7(3), 135–161 (2003).
Zhu, W., Mou, J. & Benyoucef, M. Exploring purchase intention in cross-border E-commerce: A three stage model. J. Retail. Consum. Serv. 51, 320–330 (2019).
Kim, J., Giroux, M. & Lee, J. C. When do you trust AI? The effect of number presentation detail on consumer trust and acceptance of AI recommendations. Psychol. Market 38(7), 1140–1155 (2021).
Ostrom, A. L., Fotheringham, D. & Bitner, M. J. Customer acceptance of AI in service encounters: Understanding antecedents and consequences. Handb. Serv. Sci. II, 77–103 (2019).
Qalati, S. A. et al. Effects of perceived service quality, website quality, and reputation on purchase intention: The mediating and moderating roles of trust and perceived risk in online shop. Cogent Bus. Manag. 8(1), 1869363 (2021).
Zhao, Y., Wang, L., Tang, H. & Zhang, Y. Electronic word-of-mouth and consumer purchase intentions in social e-commerce. Electron. Commer. R. A. 41, 100980 (2020).
Shin, D. The effects of explainability and causability on perception, trust, and acceptance: Implications for explainable AI. Int. J. Hum-Comput. St. 146, 102551 (2021).
Sánchez, J., Callarisa, L., Rodriguez, R. M. & Moliner, M. A. Perceived value of the purchase of a tourism product. Tourism Manage. 27(3), 394–409 (2006).
Dabbous, A. & Barakat, K. A. Bridging the online offline gap: Assessing the impact of brands’ social network content quality on brand awareness and purchase intention. J. Retail. Consum. Serv. 53, 101966 (2020).
Low, W. S., Lee, J. D. & Cheng, S. M. The link between customer satisfaction and price sensitivity: An investigation of retailing industry in Taiwan. J. Retail. Consum. Serv. 20(1), 1–10 (2013).
Watanabe, E. A. D. M., Alfinito, S., Curvelo, I. C. G. & Hamza, K. M. Perceived value, trust and purchase intention of organic food: A study with Brazilian consumers. Brit. Food J. 122(4), 1070–1184 (2020).
Heine, S. J. & Buchtel, E. E. Personality: The universal and the culturally specific. Annu. Rev. Psychol. 60(1), 369–394 (2009).
Scott, S. G. & Bruce, R. A. Determinants of innovative behavior: A path model of individual innovation in the workplace. Acad. Manage. J. 37(3), 580–607 (1994).
Podsakoff, P. M. & Organ, D. W. Self-reports in organizational research: Problems and prospects. J. Manage. 12(4), 531–544 (1986).
Spiliotopoulou, G. Reliability reconsidered: Cronbach’s alpha and paediatric assessment in occupational therapy. Aust. Occup. Ther. J. 56(3), 150–155 (2009).
Fornell, C. & Larcker, D. F. Evaluating structural equation models with unobservable variables and measurement error. J. Marketing Res. 18(1), 39–50 (1981).
Wang, S. F. & Chen, C. C. Exploring designer trust in artificial intelligence-generated content: TAM/TPB model study. Appl. Sci. 14(16), 6902 (2024).
Burhan, N. A. S. et al. Materialism and Big-Five personality traits shaping low-income university students’ compulsive online-buying behavior. Pertanika J. Soc. Sci. Hum. 30(4), 1931–1953 (2022).
Hayes, A. F. Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Commun. Monogr. 76(4), 408–420 (2009).
Thakur, R. & Srivastava, M. Adoption readiness, personal innovativeness, perceived risk and usage intention across customer groups for mobile payment services in India. Internet Res. 24(3), 369–392 (2014).
Author information
Authors and Affiliations
Contributions
L.L.H.: Conceptualization, Formal analysis, Investigation, Writing—original draft, Writing—review and editing. M.W.C.: Methodology, Resources, Supervision, Writing—review and editing. N.D.: Data curation, Conceptualization, Validation, Visualization. All authors reviewed the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Hu, L., Chen, M. & Ding, N. Factors influencing digital media designers’ subscription to premium versions of AI drawing tools through a mixed methods study. Sci Rep 15, 15994 (2025). https://doi.org/10.1038/s41598-025-99924-7
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-025-99924-7
