[Link].

com/scientificreports

OPEN Examining the impact of big five

personality traits on generation
Z designers’ subscription to paid
AI drawing tools using SEM and
FsQCA
Ning Ding, Maowei Chen & Liling Hu
With the rapid advancement of artificial intelligence (AI) technologies, AI has evolved from an
exploratory technology into a critical component of the design process. This study develops an
extended model utilizing the Big Five Personality (BFP) framework, the Technology Acceptance
Model (TAM), and Perceived Risk Theory (PRT) to examine the influence of BFP traits on Generation
Z designers’ willingness to subscribe to paid AI drawing tools. Structural equation modeling (SEM)
and fuzzy-set qualitative comparative analysis (fsQCA) were employed to analyze data from 477
valid responses. The SEM results demonstrate that Openness (OPE) and Extraversion (EXT) positively
affect Perceived Ease of Use (PEU) and Willingness to Pay for Subscription (WPS), while negatively
influencing Perceived Risk (PR). EXT and Agreeableness (AGR) enhance Perceived Usefulness (PU).
Neuroticism (NEU) adversely affects PU, PEU, and WPS but enhances PR. Moreover, WPS is positively
affected by PEU and PU, yet negatively by PR, with PEU, PU, and PR acting as mediators in certain
paths. The fsQCA findings exhibit the complex interplay of BFP traits, revealing four configurations
that influence WPS among Generation Z designers, distinctly contrasting with the SEM results. This
research systematically explores for the first time the impact of personality traits on technology
acceptance behaviors among Generation Z designers, contributing theoretical developments and
fresh insights into personality psychology and technology acceptance studies. It also offers practical
implications for technology development, marketing, education, and human resource management.

Keywords Generation Z designers, AI drawing tools, Willingness to pay for subscriptions, Complexity
theory, Big five personality, TAM

With the rapid development of AI technology, the creative industry is undergoing unprecedented changes,
drawing increasing attention from the academic community regarding the impact of AI on the creative field1.
In the realm of art and design, AI has evolved from an exploratory technology into a critical component of
the design process2. Recently, paid AI drawing tools such as MidJourney, Leonardo AI, and others have
entered the market3. These tools integrate AI technology into traditional design systems4, analyzing user input
“prompts” through language-to-visual models to generate high-quality, varied visual content5. They provide
designers with unprecedented freedom and creativity during the conceptual design stage, enhancing both
design efficiency and quality6. Compared to basic free versions, paid subscription services typically offer richer
features, a more convenient user experience, and superior output quality7. Moreover, paid subscriptions not
only provide developers with a continuous and stable income source but also permit users to access timely
high-quality product updates and feature enhancements8. However, despite the significant potential of paid AI-
powered drawing tools in enhancing design efficiency and fostering creativity, their widespread adoption still
faces several challenges9. First, designers exhibit varying levels of acceptance toward AI drawing tools. Some may
adopt a cautious stance due to factors such as technological complexity, learning costs, or perceived risks, which
could hinder actual implementation10. Second, concerns regarding the acceptance of the paid model persist,
particularly in terms of balancing perceived value and cost-effectiveness, which may influence purchasing

Department of Global Convergence, Kangwon National University, Room 307, Building 5 (Natural Sciences),
Chuncheon 24341, Kangwon-do, Republic of Korea. email: muwi@[Link]

Scientific Reports | (2025) 15:17587 | [Link] 1

[Link]/scientificreports/

decisions11. Even more critically, as a professional group highly dependent on creativity and personalized
expression, designers’ personality traits may play a crucial role in the technology adoption process9. These traits
can shape their perceptions, attitudes, and adoption behaviors toward paid AI tools, further exacerbating the
divergence in needs and acceptance among different user groups.
A “generation” is defined as “a group of individuals born during the same historical period and sharing similar
cultural backgrounds"12. While there is no universally agreed-upon time span for Generation Z, a substantial
body of research defines this cohort as individuals born between 1995 and 201213,14. This generation has been
labeled “digital natives” due to their upbringing in an era of rapid technological advancement15. According to
a recent report by Udemy, Generation Z is projected to make up 27% of the global workforce by 2025 and is
expected to become the largest workforce group by 203516. Similarly, the “2024 Work Trends Index Annual
Report,” jointly published by Microsoft and LinkedIn, surveyed about 31,000 individuals from 31 countries and
regions. The report revealed that 85% of Generation Z respondents use AI in their work, positioning them as
the generation with the highest AI adoption rate across all age groups. Furthermore, the report underscores that
71% of leaders favor hiring candidates with AI skills over those with traditional work experience17. Generation
Z exhibits unique characteristics in technology acceptance and usage behavior. They not only maintain a
positive outlook toward emerging technologies but also show a heightened level of acceptance and curiosity
for exploration18,19. Additionally, their distinctive personality traits significantly influence their behavior and
decision-making processes20,21, further shaping their adoption and use of emerging technologies.
In this context, the Big Five Personality model, commonly utilized as a theoretical framework in psychology22,
provides a multidimensional approach to understanding individual characteristics and behaviors21. Prior
studies have investigated the connection between the Big Five Personality and consumers’ willingness to
pay23–25 as well as the adoption of emerging technologies26–28, emphasizing the crucial influence of personality
traits on technology acceptance and marketing. Nevertheless, research into the association between the Big
Five Personality and Generation Z designers is notably limited, especially in terms of how personality traits
affect their selection of AI-based drawing tools. Empirical studies in this domain are nearly nonexistent.
Furthermore, although previous research has extensively examined the market potential of AI tools5,29 and
their user acceptance2,6,30, investigations focusing on users’ Willingness to Pay for Subscriptions for such tools
are still emerging. Consequently, this study aims to systematically explore how the Big Five personality traits
influence Generation Z designers’ willingness to subscribe to paid AI-based drawing tools and to provide an in-
depth analysis of their decision-making mechanisms, thereby addressing existing research gaps in this domain.
Specifically, this study investigates the following key research questions:
RQ1. How do the personality traits of Generation Z designers influence their Perceived Ease of Use, Perceived
Usefulness, and Perceived Risk of paid AI-based drawing tools?
RQ2. Which personality traits significantly predict Generation Z designers’ willingness to subscribe to AI-
based drawing tools? What are the underlying mechanisms and impact pathways?
RQ3. Do Perceived Ease of Use, Perceived Usefulness, and Perceived Risk mediate the relationship between
personality traits and willingness to subscribe?
RQ4. In complex contexts, how do the Big Five personality traits interact through different configurational
patterns to jointly influence Generation Z designers’ willingness to subscribe?
Based on 477 valid samples, this study conducted a thorough analysis of how Generation Z designers’
personality traits influence their Willingness to Pay for Subscriptions. Utilizing the Big Five Personality model,
TAM, and PRT, an extended model was developed to measure the causal relationships between the Big Five
Personality traits and various key variables using SEM31. Additionally, by applying complexity theory, fsQCA
was used to explore how the interplay of different personality traits in complex contexts collectively affects the
Willingness to Pay for Subscriptions of Generation Z designers32.
This study significantly contributes to existing literature by being the first to examine designer personality
traits’ impact on the TAM, using complexity theory to analyze the intricate interactions between various
personality traits and designers’ Willingness to Pay for Subscriptions. It responds to academic calls for more
research into personality traits’ role in adopting emerging technologies33,34, thus addressing a critical gap in the
relevant literature. Moreover, through exploratory analysis, it identifies the key mediating roles of Perceived Ease
of Use, Perceived Usefulness, and Perceived Risk as variables linking individual traits and behavioral decision-
making, providing essential empirical evidence for future studies on how individual traits affect behavioral
decisions through perceptual factors. Finally, this study provides practical strategic guidance and valuable
insights for technology development, marketing, education, and human resource management.
The structure of this paper is arranged as follows: The second section covers the literature review and
elaborates on the development of the research model and hypotheses. The third section provides a detailed
introduction to the research methodology, including questionnaire design, data collection procedures, and
relevant ethical approval and informed consent considerations. The fourth section presents the results of data
analysis at various stages. The fifth section conducts an in-depth discussion of the research findings, focusing
on their theoretical and managerial implications, while also addressing the study’s limitations and proposing
directions for future research.

Literature review
Underpinning theory
The Technology Acceptance Model (TAM)35, derived from the Theory of Reasoned Action proposed by
Fishbein and Ajzen (1975)36, has become one of the predominant theoretical frameworks for understanding
the adoption and use of information technology37. TAM is frequently used to analyze user adoption behaviors
toward new technologies10 and has proven widely applicable across different technologies and user groups38.
In the context of rapid advancements in AI, TAM’s applicability has been further broadened and enhanced10.

Scientific Reports | (2025) 15:17587 | [Link] 2

[Link]/scientificreports/

However, despite the strong theoretical value of the TAM in explaining users’ technology acceptance behavior,
its relatively simplified framework may lead to certain limitations in predicting and explaining technology
adoption behavior39. For example, Kumpulainen et al. (2025)40 highlight the boundary conditions of the
TAM across different contexts, suggesting that integrating additional theoretical frameworks is necessary to
comprehensively address socio-technical factors beyond behavioral intention. This approach helps to mitigate
the limitations of TAM and enhance its explanatory power. Furthermore, Symasek et al. (2025)41 emphasize that
TAM largely overlooks the impact of individual differences on users’ behavioral intentions. Existing research
has demonstrated that individual differences may play a critical role in the process of technology acceptance
and adoption. In recent years, a substantial number of TAM-based studies have improved the model’s overall
explanatory power and theoretical foundation by incorporating additional variables or by integrating it with
other theories42. For example, Wang and Chen (2024)10 integrated the TAM with the Theory of Planned
Behavior to develop a new theoretical framework that explores designers’ trust in and adoption of AI-generated
content. Similarly, Al-Okaily (2025)43 extended the TAM model by incorporating environmental factors, social
influence, and security-related variables alongside the original constructs of PU and PEU, offering new insights
into the application of AI in education. These studies further validate the effectiveness of TAM and its extended
models across various technology acceptance contexts, highlighting the model’ s continued relevance for future
information technology research44.
In real-world contexts, individuals’ personality traits significantly influence their perceptions of and behaviors
toward technology45,46. Therefore, incorporating the BFP framework to account for individual differences
can enhance the explanatory power of the TAM. The BFP model is a comprehensive and widely recognized
framework for classifying personality traits22. It categorizes personality into five dimensions: OPE, CON, AGR,
EXT, and NEU24,25, and has significantly influenced various fields. In organizational behavior, it is extensively
used for recruitment and personnel selection47,48, performance evaluation49,50, and leadership research51,52. In
environmental studies, it is often applied to explore attitudes and behaviors toward sustainable development53,54.
Moreover, the BFP model contributes notably to educational and occupational psychology by predicting
academic performance55,56 and influencing career choices57,58. The BFP model is particularly significant in
understanding individuals’ perceptions of technology value and their corresponding behaviors59. Numerous
studies have examined BFP as a predictive factor for technology adoption and usage. Malik & Singh (2022)60
explored the impact of BFP on the adoption and continued use of green banking channels, demonstrating that
AGR and EXT positively influenced PU and PEU, while CON and OPE positively affected only PEU. Wang et
al. (2024)28 analyzed BFP’s influence on the willingness to use virtual reality learning systems, finding positive
correlations between NEU and PU and negative correlations between CON and PU. Kumar et al. (2024)27
investigated the role of BFP in users’ willingness to adopt the metaverse and found that EXT, AGR, and NEU
significantly influenced willingness, whereas CON and OPE had no significant effect. Similarly, Kannen et al.
(2024)45 studied BFP’s impact on users’ willingness to pay for social media services, identifying no significant
correlation between NEU and willingness to pay, while other traits showed positive correlations. These findings
underscore that the influence of BFP varies across different contexts, highlighting the need for further in-depth
and comprehensive research to explore its multifaceted roles.
PRT was initially introduced by Bauer (1967)61, who defined PR as an individual’s subjective judgment or
value assessment rooted in multiple objective threats in uncertain situations. This theory is extensively applied
in technology acceptance research to elucidate individuals’ cognition and decision-making behaviors when
confronted with potential risks62. The original TAM does not fully account for the impact of PR on technology
acceptance. However, in many technological application contexts—such as online payment systems, AI-driven
solutions, and blockchain—users’ perceptions of risks related to privacy, security, and trust often serve as critical
determinants of adoption intention63,64. Therefore, integrating the PRT can enhance the model’s applicability
to real-world scenarios. Recent studies have incorporated PR into the TAM framework6,65,66, demonstrating its
effectiveness and reliability in boosting TAM’s explanatory power. A substantial body of research indicates that
high levels of PR significantly reduce individuals’ willingness to accept technology2,6,10. However, in recent years,
some scholars have begun to explore the concept of AI awareness, defining it as employees’ risk perceptions
regarding the potential of AI technologies (such as robots, algorithms, and intelligent systems) to replace their
jobs and impact career development67,68. Some emerging studies suggest that this risk perception exhibits a dual
effect—while it may undermine employees’ proactivity and innovative behaviors, under certain conditions, it can
also stimulate adaptive adjustments, encouraging them to actively cope with technological change69,70. However,
it should be noted that the focus of this study is distinct from the social implications of AI awareness. Instead,
it concentrates on the potential risks inherent in AI drawing tools themselves, including considerations related
to privacy, security, and other related concerns64. Therefore, this study does not address the dual effect of AI
awareness on technology acceptance but instead focuses on the risk perceptions of AI drawing tools in specific
usage contexts and their underlying mechanisms. Moreover, existing studies emphasize the significant impact
of individual personality traits on TAM constructs26,28,71, revealing considerable variations in the perception of
technological risks among individuals with diverse personality traits72.
In summary, this study integrates the TAM framework with the BFP model and PRT to develop a new extended
model (see Fig. 1), providing more profound insights than using TAM alone. As it examines the correlation
between BFP and WPS, it adheres to the principle of Occam’s Razor—“entities should not be multiplied beyond
necessity”—(McSharry & Mawejje, 2024)73, thus keeping the model complexity within necessary limits. A
simplified model not only enhances interpretative clarity but also mitigates the risk of SEM overfitting74. To
prevent unnecessary complexity and assumptions, the study excludes the classical relationship between PEU and
PU, along with non-essential relations between PEU, PU, and PR from the proposed model.
Notably, different personality traits may exhibit nonlinear and dynamic interactions58,64, which go beyond the
explanatory capacity of the traditional TAM. Therefore, complexity theory offers a more insightful perspective

Scientific Reports | (2025) 15:17587 | [Link] 3

[Link]/scientificreports/

Fig. 1. Research model.

for further advancing this study. The origins of complexity theory can be traced back to the fields of science
and mathematics75, with significant contributions from early system sciences such as cybernetics, information
theory, and general systems theory76. Over time, complexity theory evolved into an independent area of study,
distinct from other system theories77. As a paradigm shift and a new framework relative to traditional science78,
complexity theory seeks to explore behavioral patterns and dynamic relationships within complex systems,
particularly how systems experience nonlinear, transformative changes rather than linear, incremental ones79.
In other words, complexity theory not only focuses on the individual components of a system but also explores
how these components interact to form a novel entity, emphasizing that the whole is greater than the mere
sum of its parts and interactions77. In recent years, this theory has found widespread applications in fields such
as management80,81, sociology82,83, and behavioral sciences84,85. With the increasing prevalence of complex
challenges across various disciplines, the significance of complexity theory in both academic and practical
domains has become increasingly evident77. This theory not only equips researchers with diverse tools, methods,
and techniques to address complex systems but also fosters a new way of thinking, encouraging individuals to
rethink, study, and manage the complexities of an ever-changing world86. In this study, complexity theory helps
to understand how different personality traits interact during the technology acceptance process and shape
individual decision-making. This approach provides a more precise explanatory framework and offers insights
with greater practical relevance.

Hypotheses development
OPE is closely associated with creativity, curiosity, and imagination. Individuals with high OPE exhibit strong
abstract thinking skills, appreciation for diversity, and enjoyment of unique experiences87. They generally
maintain positive attitudes towards novel ideas and experiences and demonstrate a readiness to embrace
change and innovation88. These individuals often pursue opportunities to acquire new skills via technology,
believing that it can enhance their performance89. They are also more predisposed to accept and tolerate the
risks associated with new opportunities, often viewing these risks as positive influences90. Previous research has
shown that OPE has a significant positive influence on PEU and PU26,71 and a significant negative influence on
PR63. Based on this discussion, the following hypotheses are proposed:

H1a OPE is positively associated with PEU.

H1b OPE is positively associated with PU.

H1c OPE is negatively associated with PR.

CON emphasizes prudence, diligence, responsibility, goal orientation, and self-discipline. Individuals with
high levels of CON typically exhibit strong self-management abilities, engage in planned behavior23, and are
willing to adopt technology to enhance their efficiency, performance, and productivity91. However, they evaluate
whether technological innovations provide meaningful career advancement opportunities before adopting
them92. Additionally, individuals with high CON generally possess strong achievement motivation and a sense
of commitment, demonstrating a willingness to take risks to achieve their goals90. Previous studies have shown
that CON significantly enhances PEU and PU91,93 while exerting a significant negative effect on PR94. Based on
this discussion, the following hypotheses are proposed:

Scientific Reports | (2025) 15:17587 | [Link] 4

[Link]/scientificreports/

H2a CON is positively associated with PEU.

H2b CON is positively associated with PU.

H2c CON is negatively associated with PR.

EXT primarily reflects individuals’ levels of socialization and vitality. Individuals with high levels of EXT are
more likely to engage in social interactions, exhibit high energy, and thrive in group environments23. They excel
at socializing and enjoy attracting attention, using cutting-edge technology as a means to enhance their image95.
Driven by intrinsic motivations, individuals with high EXT are inclined to become risk-seekers, often viewing
risk as a positive opportunity90. Previous studies have shown that as EXT levels increase, individuals’ PEU and
PU also increase60,96, while PR shows a decreasing trend97. Based on this discussion, the following hypotheses
are proposed:

H3a EXT is positively associated with PEU.

H3b EXT is positively associated with PU.

H3c EXT is negatively associated with PR.

AGR is associated with traits such as cooperativeness, compassion, and concern for others. Individuals with
higher levels of AGR tend to prioritize harmonious interactions, display empathy, and value the feelings of
others98. Furthermore, highly agreeable individuals are more likely to enjoy using new technologies and share
their positive experiences with others28. In various contexts, these individuals often demonstrate proactive and
constructive behaviors, viewing risks as potential opportunities or as beneficial factors for others90. Previous
studies have shown that higher levels of AGR are positively associated with PEU and PU71,96, while being
negatively correlated with PR99. Based on the above discussion, the following hypotheses are proposed:

H4a AGR is positively associated with PEU.

H4b AGR is positively associated with PU.

H4c AGR is negatively associated with PR.

NEU reflects an individual’s emotional stability and capacity to handle stress. Individuals with high levels of NEU
are often more susceptible to anxiety, frustration, and emotional fluctuations100. They tend to feel apprehensive
about potential losses and are generally hesitant to embrace new experiences23. Consequently, they typically
adopt a cautious approach toward new technologies, often seeking additional information or assurances before
making decisions23. Moreover, individuals with elevated NEU are less likely to adopt new technologies59 due to
their heightened sensitivity to risks and potential negative outcomes associated with technology use23. Unlike
other personality traits, prior research has consistently shown that higher levels of NEU are associated with lower
PEU and PU26,60, while PR tends to increase significantly64. Based on this discussion, the following hypotheses
are proposed:

H5a NEU is negatively associated with PEU.

H5b NEU is negatively associated with PU.

H5c NEU is positively associated with PR.

Existing research suggests a direct correlation between an individual’s personality traits and their technology
acceptance behavior. Specifically, OPE, CON, EXT, and AGR are generally considered to have positive
correlations with technology adoption intention45,46. This implies that the more pronounced these personality
traits are, the more likely an individual’s acceptance of technology will be positive. Conversely, NEU has been
found to negatively correlate with technology adoption intention27,101, indicating that individuals with high
levels of NEU may exhibit greater resistance or concern toward technology use. Based on this discussion, the
following hypotheses are proposed:

H1 d OPE is positively associated with WPS.

H2 d CON is positively associated with WPS.

H3 d EXT is positively associated with WPS.

H4 d AGR is positively associated with WPS.

H5 d NEU is negatively associated with WPS.

Scientific Reports | (2025) 15:17587 | [Link] 5

[Link]/scientificreports/

As critical factors influencing system adoption and use, PEU and PU have been consistently validated for their
robustness and reliability across numerous studies37,102. Within the TAM, PU and PEU are significant variables
that exhibit a strong positive correlation with self-predicted future usage behavior, with PU showing a stronger
correlation with behavioral intention than PEU35. This foundational relationship, revealed by the TAM, has
been widely supported and confirmed in extensive research66,103. Additionally, the negative impact of PR on
technology adoption intention has been well-documented6,65. Based on the above discussion, the following
hypotheses are proposed:

H6 PEU is positively associated with WPS.

H7 PU is positively associated with WPS.

H8 PR is negatively associated with WPS.

To further explore the potential roles of PEU, PU, and PR, this study will examine their mediating effects
between BFP and WPS during the data analysis phase. Although there is currently a lack of empirical studies
directly supporting this pathway, existing research on the TAM and its extended models suggests that PU and
PEU significantly influence technology adoption behaviors66. Meanwhile, PR is commonly regarded as a critical
inhibitor in decision-making processes6. Therefore, this study speculates that these perceptual factors may
mediate the relationship between BFP and WPS. An exploratory analysis will be conducted to examine this
speculation, providing deeper empirical insights for future research.

Methodology
This study employs SEM and fsQCA for data analysis. SEM is an advanced statistical modeling approach that
has become a crucial tool for analyzing multivariate relationships due to its strong estimation capabilities,
modeling flexibility, and wide-ranging applications104. Currently, SEM is primarily categorized into two types:
partial least squares SEM (PLS-SEM) and covariance-based SEM (CB-SEM)105. Among these, CB-SEM is
commonly implemented using software such as AMOS, EQS, LISREL, and MPlus to validate and test research
hypotheses106. This study adopts CB-SEM, primarily due to its advantages in theoretical hypothesis validation,
handling complex multivariate relationships, statistical testing, and model fit evaluation107.
fsQCA, first introduced by Charles Ragin108, is an empirical method based on complexity theory. It follows key
principles such as conjunctural causation, equifinality, and causal asymmetry to support inductive research109.
fsQCA goes beyond the linear models of traditional causal inference methods, such as correlation, regression,
and SEM, thus enabling the examination of the complex interplay of multiple conditions on outcomes110. This
method is especially adept at thoroughly exploring particular social science phenomena111. As fsQCA has become
widely accepted and continuously advanced110, Pappas & Woodside (2021)32 issued a significant technical note
that expounded its core principles and offered detailed guidelines for its application in research settings.
The integration of fsQCA with SEM has become a significant focus in academia, with extensive applications
in areas such as consumer behavior84,112,113 and technology acceptance37,114. The combined application of fsQCA
and SEM has proven its viability and effectiveness in exploring complex social phenomena, thus providing
researchers with a powerful tool to delineate intricate causal networks. In addition to the core analytical methods,
this study employed SPSS 26.0 for frequency analysis, Cronbach’s a reliability testing, and Common Method Bias
(CMB) testing. Furthermore, the PROCESS macro was utilized to further investigate potential mediating effects.

Questionnaire design
All measurement scales in this study were adapted from established and validated scales. To ensure the accuracy
and reliability of these scales, experts from the fields of personality psychology, technology acceptance, and
design were invited to rigorously review and refine them. As a result, the finalized scales comprise nine variables
and 33 items (see Appendix A). The experts unanimously agreed that the revised scales meet high academic
standards in terms of structural validity and overall quality.

Research participants and data collection

The questionnaire in this study is divided into two sections. The first section collects basic demographic
information of the respondents, while classification of professional backgrounds is based on the studies by Li
(2024)6 and Wang et al. (2024)10. The second section includes nine constructs from the research model, which
are measured on a 5-point Likert scale, with ‘1’ representing ‘strongly disagree’ and ‘5’ representing ‘strongly
agree’.
The formal data collection period lasted from October 9, 2024, to October 30, 2024. This study focuses on
paid AI image generation tools such as MidJourney and Leonardo AI, which exhibit a high degree of similarity
in terms of functionality, pricing models, and accessibility. These tools are all based on text-to-image generation
technology6, and adopt either a subscription-based or pay-per-use model115. Given the absence of significant
differences in their core functionalities, users’ willingness to pay may be more influenced by their individual
personality traits9. Therefore, this study treats AI image generation tools as a unified category and does not
distinguish between specific tool sources during sample recruitment, thereby enhancing the generalizability and
external validity of the research. A purposive sampling method was used to recruit relevant respondents, targeting
Gen Z design professionals aged 18–29 with experience using paid AI drawing tools13,14. Purposeful sampling is
primarily based on a targeted sample selection strategy, enabling researchers to obtain more representative data
by clearly defining research questions and target audiences, thereby enhancing the validity and relevance of the
data116. This method not only improves the efficiency of data collection but is particularly advantageous in this

Scientific Reports | (2025) 15:17587 | [Link] 6

[Link]/scientificreports/

study, which focuses on Generation Z designers with age and experience constraints. It allows for the precise
identification of respondents who meet the research criteria, ensuring the specificity and reliability of the data.
Questionnaires were distributed via instant messaging and community communication platforms, particularly
Discord, where monetary incentives were provided to eligible participants. Discord was chosen as the survey
platform for two reasons: (1) well-known AI drawing tools like MidJourney as well as [Link] are accessible
and operable via Discord accounts, thereby attracting a significant community of design professionals to the
platform; (2) Discord hosts various official AI drawing tool communities, creator communities, and technical
exchange groups, offering a strategic advantage in accurately targeting the respondent group. This approach
enhances the representativeness of the sample and the relevance of the survey.
Hair et al. (2010)117 recommend associating each variable and its corresponding items with 10–15 participants.
Accordingly, the current study, comprising 9 constructs and 33 items, requires an effective sample size of 330–495.
Taking into account a 10% attrition rate and invalid responses, the estimated sample size ranges from 367 to 550.
The minimum sample size was calculated with the G*Power statistical test and empirical rules, setting parameters
as f2 = 0.15, α = 0.05, and Power = 0.80, and a calculated minimum of 160 was required118. The questionnaire
was circulated through the professional online survey platform “Wenjuanxing” to 540 designers experienced
in AI drawing tools subscriptions. All participants were informed about the anonymity and confidentiality of
the survey and assured that responses were subjective. A total of 512 questionnaires were returned, reflecting a
response rate of 94.8%. After removing 35 invalid or incomplete responses, 477 valid questionnaires remained,
meeting the sample size recommendations of Hair et al. (2010)117 and Faul et al. (2009)118.

Ethical approval and informed consent

This study has been approved by the Department of Global Integration at Kangwon National University, and all
research methods strictly follow relevant guidelines and regulations. Respondents participating in the online
survey are required to read the ethical statement at the top of the questionnaire before completing it. They may
only proceed with the survey after clicking to agree. The participants in this study were all adults, and informed
consent was obtained from all.

Results
Respondent profile and CMB test
The demographic characteristics of respondents are detailed in Table 1. There were 237 male (49.7%) and
240 female participants (50.3%), indicating a slightly higher female presence. Most respondents, numbering
95 (19.9%), came from animation design. Regarding experiences with paid AI drawing tool subscriptions, the
majority, amounting to 166 users (34.8%), reported durations of 1–6 months.
Since all data in this study were sourced from respondents’ self-reports, there is a potential risk of CMB119.
To assess this, the Harman single-factor test was utilized120. Nine non-rotated common factors with eigenvalues
greater than 1 were identified, explaining 74.224% of the total variance. The first factor accounted for only
9.199% of the variance, remaining below the critical threshold of 40%, indicating the absence of significant CMB
issues in this study according to Podsakoff & Organ (1986)121.

Results of measurement model

This study employed SPSS 26.0 to calculate the Cronbach’s alpha coefficient, assessing the reliability of the
measurement items. Typically, scales with Cronbach’s alpha coefficients between 0 and 1, with values above 0.7,
are deemed reliable122. The results (Table 2) show that the Cronbach’s alpha coefficients for each construct were
above 0.7, indicating high internal consistency and reliability.

Items Categories Frequency %

Male 237 49.7
Gender
Female 240 50.3
Visual communication design 79 16.6
Environmental design 45 9.4
Product design 92 19.3
Animation design 95 19.9
.Professional background Industrial design 20 4.2
Fashion design 46 9.6
Game design 58 12.2
Architectural design 24 5.0
Others 18 3.8
< 1 month 60 12.6
1–6 months 166 34.8
Experience with paid AI drawing tools subscriptions
6 months–1 year 162 34.0
> 1 year 89 18.7

Table 1. Demographic characteristics of the respondents (N = 477).

Scientific Reports | (2025) 15:17587 | [Link] 7

[Link]/scientificreports/

Constructs Mean SD Factor loadings a AVE CR

OPE1 3.36 1.057 0.816
OPE2 3.37 1.036 0.806
OPE 0.870 0.627 0.870
OPE3 3.31 1.044 0.796
OPE4 3.34 1.044 0.748
CON1 3.54 1.174 0.723
CON2 3.58 1.129 0.804
CON 0.862 0.611 0.863
CON3 3.55 1.125 0.801
CON4 3.57 1.140 0.796
EXT1 3.42 0.970 0.755
EXT2 3.39 0.987 0.854
EXT 0.876 0.639 0.876
EXT3 3.38 0.990 0.811
EXT4 3.38 0.987 0.775
AGR1 3.36 1.162 0.807
AGR2 3.39 1.156 0.792
AGR 0.885 0.657 0.885
AGR3 3.40 1.144 0.815
AGR4 3.39 1.163 0.828
NEU1 2.43 0.978 0.756
NEU2 2.39 0.980 0.868
NEU 0.867 0.625 0.869
NEU3 2.41 1.004 0.761
NEU4 2.41 1.008 0.773
PEU1 3.79 0.932 0.727
PEU PEU2 3.76 0.926 0.841 0.832 0.627 0.834
PEU3 3.75 0.964 0.804
PU1 3.57 0.976 0.840
PU PU2 3.60 1.024 0.875 0.896 0.744 0.897
PU3 3.60 0.990 0.873
PR1 2.55 1.083 0.805
PR PR2 2.54 1.097 0.800 0.837 0.632 0.837
PR3 2.52 1.109 0.779
WPS1 3.56 1.008 0.810
WPS2 3.57 0.993 0.762
WPS 0.861 0.608 0.861
WPS3 3.61 1.006 0.765
WPS4 3.52 1.024 0.781

Table 2. The reliability and validity of the measurement tool. Note. SD: Standard deviation, a: Cronbach’s
alpha, AVE: Average variance extracted, CR: Composite reliability.

WPS PR PU PEU NEU AGR EXT CON OPE

WPS 0.780
PR −0.435 0.795
PU 0.423 −0.359 0.863
PEU 0.438 −0.268 0.344 0.792
NEU −0.438 0.400 −0.354 −0.301 0.791
AGR 0.128 −0.120 0.356 0.073 −0.092 0.811
EXT 0.425 −0.438 0.326 0.314 −0.258 0.075 0.799
CON 0.030 −0.062 0.047 0.062 −0.083 −0.011 0.047 0.782
OPE 0.368 −0.366 0.186 0.354 −0.258 0.081 0.267 0.049 0.792

Table 3. Discriminant validity analysis. Note. Bold values represent the square root of the AVE.

CFA was conducted using Amos 23.0. CFA provides rigorous support and enhances validity assessments
while mitigating potential biases from intuitive methods123. The internal consistency and convergent validity of
constructs were assessed using composite reliability (CR) and average variance extracted (AVE). These metrics
(Table 3) reveal that the AVE values for each construct exceeded 0.5 and the CR values exceeded 0.7, indicating
strong convergent validity124. Discriminant validity was assessed using Fornell & Larcker’s (1981)124 criterion,

Scientific Reports | (2025) 15:17587 | [Link] 8

[Link]/scientificreports/

Index Desired value CFA model SEM model

χ²/df <3 1.450 1.615
GFI >0.9 0.923 0.911
AGFI >0.8 0.906 0.895
RMSEA <0.08 0.031 0.036
IFI >0.9 0.975 0.965
CFI >0.9 0.975 0.965
TLI >0.9 0.971 0.961

Table 4. Model fit test. Note. χ²: Chi-square, df: Degrees of freedom, GFI: Goodness-of-fit index, AGFI:
Adjusted goodness-of-fit index, RMSEA: Root mean square error of approximation, IFI: Incremental fit index,
CFI: Comparative fit index, TLI: Tucker-lewis index.

Hypotheses Path relationships B b C.R. p Results

H1a OPE→PEU 0.201 0.266 4.801 *** Supported
H1b OPE→PU 0.055 0.057 1.152 0.249 Rejected
H1c OPE→PR −0.223 −0.236 −4.505 *** Supported
H2a CON→PEU 0.021 0.029 0.575 0.566 Rejected
H2b CON→PU 0.019 0.020 0.432 0.665 Rejected
H2c CON→PR −0.019 −0.021 −0.434 0.665 Rejected
H3a EXT→PEU 0.188 0.220 4.036 *** Supported
H3b EXT→PU 0.271 0.248 4.974 *** Supported
H3c EXT→PR −0.362 −0.339 −6.301 *** Supported
H4a AGR→PEU 0.023 0.034 0.679 0.497 Rejected
H4b AGR→PU 0.284 0.326 6.749 *** Supported
H4c AGR→PR −0.058 −0.069 −1.404 0.160 Rejected
H5a NEU→PEU −0.171 −0.203 −3.752 *** Supported
H5b NEU→PU −0.297 −0.276 −5.493 *** Supported
H5c NEU→PR 0.303 0.289 5.440 *** Supported
H1 d OPE→WPS 0.124 0.145 2.737 0.006 Supported
H2 d CON→WPS −0.022 −0.027 −0.594 0.553 Rejected
H3 d EXT→WPS 0.184 0.190 3.421 *** Supported
H4 d AGR→WPS 0.001 0.001 0.024 0.981 Rejected
H5 d NEU→WPS −0.204 −0.213 −3.869 *** Supported
H6 PEU→WPS 0.222 0.195 3.604 *** Supported
H7 PU→WPS 0.153 0.173 3.112 0.002 Supported
H8 PR→WPS −0.114 −0.125 −2.161 0.031 Supported

Table 5. Direct effects test. Note. B: Unstandardized coefficient, b: Standardized coefficient. C.R.: Critical ratio,
***p < 0.001.

wherein the square roots of the AVE for each construct were greater than their correlation coefficients with other
constructs (Table 2). This confirms robust discriminant validity of the measurement model.
According to the standards proposed by Burhan et al. (2022)24, Zhou et al. (2024)2, Qu et al. (2024)84 and
Wang & Chen (2024)10, the measurement model demonstrated an acceptable fit (see Table 4). These findings
highlight a good alignment between the measurement model and the observed data.

Results from SEM and mediation effect analysis

The goodness-of-fit and path significance of the SEM model were evaluated using AMOS 23.0. The fit indices
(Table 5) indicate a good fit. The results of the SEM path significance tests (Table 6; Fig. 2) reveal that OPE (b =
0.266, p < 0.001), EXT (b = 0.220, p < 0.001), and NEU (b = −0.203, p < 0.001) significantly influence PEU, while
CON (b = 0.029, p = 0.566) and AGR (b = 0.034, p = 0.497) do not significantly influence PEU. Thus, hypotheses
H1a, H3a, and H5a are supported, while H2a and H4a are rejected. Regarding impacts on PU, EXT (b = 0.248,
p < 0.01), AGR (b = 0.326, p < 0.001), and NEU (b = −0.276, p < 0.001) show significant impacts, while OPE
(b = 0.057, p = 0.249) and CON (b = 0.020, p = 0.665) do not. Consequently, hypotheses H3b, H4b, and H5b are
substantiated, while H1b and H2b are rejected. Regarding PR, the results indicate significant effects for OPE (b
= −0.236, p < 0.001), EXT (b = −0.339, p < 0.001), and NEU (b = 0.289, p < 0.001), while effects from CON (b
= −0.021, p = 0.665) and AGR (b = −0.069, p = 0.160) are not significant. Therefore, hypotheses H1c, H3c, and
H5c are supported, whereas H2c and H4c are rejected. In terms of WPS, the results show that OPE (b = 0.145,

Scientific Reports | (2025) 15:17587 | [Link] 9

[Link]/scientificreports/

95% CI
Paths Effect S.E. LLCI ULCI Results
A1: OPE→PEU→WPS 0.0352 0.0124 0.0135 0.0626 Supported
A2: OPE→PU→WPS −0.0035 0.0078 −0.0198 0.0118 Rejected
A3: OPE→PR→WPS 0.0230 0.0108 0.0046 0.0468 Supported
A4: CON→PEU→WPS 0.0023 0.0070 −0.0123 0.0168 Rejected
A5: CON→PU→WPS 0.0016 0.0062 −0.0108 0.0143 Rejected
A6: CON→PR→WPS −0.0002 0.0053 −0.0115 0.0108 Rejected
A7: EXT→PEU→WPS 0.0262 0.0117 0.0062 0.0521 Supported
A8: EXT→PU→WPS 0.0225 0.0110 0.0048 0.0471 Supported
A9: EXT→PR→WPS 0.0309 0.0145 0.0066 0.0635 Supported
A10: AGR→PEU→WPS −0.0040 0.0063 −0.0168 0.0084 Rejected
A11: AGR→PU→WPS 0.0367 0.0129 0.0146 0.0648 Supported
A12: AGR→PR→WPS 0.0016 0.0052 −0.0087 0.0122 Rejected
A13: NEU→PEU→WPS −0.0222 0.0110 −0.0464 −0.0035 Supported
A14: NEU→PU→WPS −0.0303 0.0113 −0.0552 −0.0108 Supported
A15: NEU→PR→WPS −0.0242 0.0111 −0.0482 −0.0051 Supported

Table 6. Indirect effects test. Note. LLCI: Lower limit confidence interval, ULCI: Upper limit confidence
interval. S.E.: Standard error.

Fig. 2. Results of model analysis.

p = 0.006), EXT (b = 0.190, p < 0.001), NEU (b = −0.213, p < 0.001), PEU (b = 0.195, p < 0.001), PU (b = 0.173, p =
0.002), and PR (b = −0.125, p = 0.031) significantly influence WPS. However, CON (b = −0.027, p = 0.553) and
AGR (b = 0.001, p = 0.981) do not show significant effects on WPS. Accordingly, hypotheses H1 d, H3 d, H5 d,
H6, H7, and H8 are supported, whereas H2 d and H4 d are rejected.
To examine whether PEU, PU, and PR mediate the impact of BFP traits on WPS, a bias-corrected percentile
bootstrap test was performed using the Process plugin in SPSS 26.0. This analysis, conducted with 5,000
resamples and a 95% confidence interval. According to Hayes (2009)125, when the confidence interval does not
include 0, it indicates a significant mediating effect. The results (Table 6) indicate that, except for paths A2, A4,

Scientific Reports | (2025) 15:17587 | [Link] 10

[Link]/scientificreports/

Fuzzy set
calibration Descriptive statistics
Before calibration FM CP FNM After calibration Mean SD Min Max
OPE 4 3.25 3 FOPE 0.464 0.444 0.001 1
CON 4 3.75 3 FCON 0.524 0.452 0.001 1
EXT 4 3.25 3 FEXT 0.491 0.438 0.001 1
AGR 4 3.25 3 FAGR 0.495 0.449 0.001 1
NEU 3 2.25 2 FNEU 0.129 0.294 0.001 1
WPS 4.25 3.75 3 FWPS 0.897 0.236 0.001 1

Table 7. Data calibration and descriptive statistics (N = 477). Note. FM: Full membership, CP: Crossover
point, FNM: Full non-membership, SD: Standard deviation, Min: Minimum, Max: Maximum.

Conditions Consistency Coverage

FOPE 0.504 0.973
~FOPE 0.515 0.862
FCON 0.539 0.922
~FCON 0.478 0.902
FEXT 0.535 0.977
~FEXT 0.483 0.851
FAGR 0.516 0.935
~FAGR 0.501 0.890
FNEU 0.109 0.760
~FNEU 0.904 0.931

Table 8. Necessity analysis of condition variables. Note: “~ ” indicates the absence of a condition.

A5, A6, A10, and A12, the upper and lower bounds of the bootstrap confidence intervals for all other paths do
not include 0, confirming significant mediating effects for these paths.

Results from FsQCA

SEM results illustrated unidirectional relationships between the variables. Nevertheless, due to the mutual
independence of the five traits in the BFP model, individuals may exhibit varying levels across each dimension58,64,
implying that the net effect of a single variable may not fully or accurately explain the research phenomenon.
Analyzing interaction effects can provide more nuanced and supplementary insights84. To comprehensively
explore the causal condition combinations influencing WPS among Generation Z designers, this study used the
fsQCA method to reveal combination patterns among different personality traits, resulting in more complex and
diverse research conclusions112.

Data calibration
Data calibration is a crucial preliminary step before implementing fsQCA126. In this study, OPE, CON, EXT,
AGR, and NEU were selected as condition variables, and WPS was designated as the outcome variable. Since all
dimensions were assessed using a 5-point Likert scale, it was necessary to calibrate the raw scale data into fuzzy
sets ranging from 0 to 142. Before calibration, the index for each dimension was derived from the average scores
of the items within that dimension114. The upper quartile (75%), median (50%), and lower quartile (25%) of
each index served as the calibration anchors for full membership, the crossover point, and full non-membership,
respectively127. Subsequently, the direct calibration method outlined by Ragin (2008)126 was implemented, and
the three anchors for each variable were entered into the Calibrate function of the fsQCA 3.0 software for data
calibration. Given that a substantial number of cases attained membership scores of exactly 0.5 after calibration,
and to circumvent the use of precise 0.5 membership scores as causal conditions126, the study adhered to the
technical guidelines of Pappas and Woodside (2021)32. A minimal increment of 0.001 was applied to causal
conditions with membership scores below 1 to ensure inclusion of all cases in the fuzzy set analysis and to
prevent data loss. The calibration results are displayed in Table 7.

Necessity analysis
After data calibration, a necessity analysis was performed to assess whether any of the five condition variables
are requisite for WPS. Based on Ragin’s (2008)126 criterion, a variable is considered necessary if its consistency
surpasses 0.9. The results, shown in Table 8, indicate that the consistency of ~ FNEU exceeds 0.9, establishing
~ FNEU as a necessary condition for WPS.

Scientific Reports | (2025) 15:17587 | [Link] 11

[Link]/scientificreports/

Solutions
Conditions S1 S2 S3 S4
FOPE ●
FCON ▲ ▲
FEXT ●
FAGR ●
FNEU 〇
Consistency 0.931 0.935 0.982 0.982
Raw coverage 0.904 0.516 0.294 0.308
Unique coverage 0.263 0.025 0.004 0.005
Solution consistency 0.923
Solution coverage 0.958

Table 9. Configurations of conditions leading to WPS. Note. ● indicates the presence of a core condition,
〇 represents the absence of a core condition, ▲ signifies the presence of a peripheral condition, and a blank
space suggests that the condition is optional or irrelevant.

Sufficiency analysis
Following the steps outlined by Ragin (2008)126, the fuzzy set [Link] was executed using fsQCA 3.0, generating
a truth table with 2k rows, where k represents the number of condition variables. Each row in the truth table
displays a unique combination of the five condition variables, along with their frequency and consistency. Given
the sample size of over 150 cases in this study, the frequency threshold was set at 332,127, and the consistency
threshold was established at 0.85 to refine the truth Table126. The fsQCA results yielded three types of solutions:
the complex solution, the intermediate solution, and the parsimonious solution. Since the intermediate solution
is considered superior in terms of both completeness and interpretability126, this study selected the intermediate
solution for final interpretation. Additionally, core and peripheral conditions within each combination were
identified by comparing the pathways in the intermediate and parsimonious solutions127.
Table 9 presents four configurations that lead to WPS, with an overall solution coverage of 0.958 and
consistency of 0.923, indicating strong combined explanatory power and consistency126. Additionally, the
consistency values of each configuration exceed 0.85, confirming that all are sufficient conditions for WPS42.
In configurations S1 and S2, ~FNEU and FAGR, respectively, function as core conditions, independent of other
conditions. S1 exhibits the highest raw coverage (0.904) and unique coverage (0.263) among all configurations,
with a consistency of 0.931, highlighting the crucial role of low NEU. Similarly, S2 demonstrates high consistency
(0.935) and raw coverage (0.516), emphasizing the positive impact of AGR on WPS. The consistency values of
configurations S3 (FOPE * FCON) and S4 (FEXT * FCON) reach the highest levels among all configurations,
both at 0.982, with relatively high raw coverages of 0.294 and 0.308, respectively. In S3 and S4, FOPE and FEXT
serve as core conditions, while FCON acts as a peripheral condition. These configurations collectively provide
a multidimensional perspective, offering deep insights into the key personality trait combinations influencing
WPS and their underlying mechanisms.

Discussion and conclusion

Discussion
This study utilized SEM and fsQCA to systematically examine the impact of BFP traits on Generation Z designers’
willingness to subscribe to paid AI drawing tools. An expanded model was constructed based on the BFP model,
the TAM, and PRT, with corresponding hypotheses proposed. Theoretical models and hypotheses were validated
and analyzed using SEM, while exploratory mediation analysis was performed using Process macros. Moreover,
the fsQCA method exposed the complex impact mechanisms of various personality trait combinations on WPS
under different configuration conditions, thus enhancing and deepening the insights obtained from SEM.
The results of the SEM and exploratory mediation analysis yield several key insights. Firstly, EXT and AGR
significantly and positively influence PU, corroborating the findings of Malik & Singh (2022)60 and Liu et al.
(2024)96. The substantial negative effect of NEU on PU supports the studies done by Srivastava et al. (2021)26
and Malik & Singh (2022)60, yet contradicts Wang et al. (2024)28. This disparity may stem from differences in
research subjects and technological environments. In this study, the learning cost and technological uncertainty
of AI tools may exacerbate anxiety in neurotic individuals, thereby reducing PU. In contrast, the VR learning
system in Wang et al. (2024)28 enhances user comfort through an immersive experience, leading to an increase
in PU. Furthermore, PU significantly mediates the relationship between EXT, AGR, NEU, and WPS. Specifically,
Generation Z designers with higher EXT are proactive in social interactions and prioritize their self-image95,
while those with higher AGR typically display gentle personalities and cooperative attitudes98. Both groups
tend to appreciate PU positively, enhancing their WPS. Conversely, Generation Z designers with high NEU
levels are more focused on potential technology risks23, leading to a reduced positive perception of PU and
subsequently lower WPS. However, the influence of OPE and CON on PU was not significant, a result that
diverges from Agyei et al. (2020)71 but aligns with Malik & Singh (2022)60. The contradictory results may stem
from the fact that OPE facilitates users’ acceptance of new technology in the mobile banking environment,
thereby enhancing PU, whereas in the context of AI drawing tools, it tends to focus more on creativity rather

Scientific Reports | (2025) 15:17587 | [Link] 12

[Link]/scientificreports/

than practicality. Similarly, CON contributes to strengthening PU in structured tasks, but in innovative tasks, its
emphasis on cautious evaluation of tool reliability may diminish the perception of PU. Secondly, OPE and EXT
significantly enhance PEU, whereas NEU negatively affects it, aligning with previous research60. Additionally,
PEU significantly mediates between OPE, EXT, NEU, and WPS. Specifically, Generation Z designers with higher
OPE levels show robust curiosity and a desire for new experiences89, and those with higher EXT demonstrate
increased confidence and optimism23. These traits positively influence their perception of PEU, thus improving
their WPS. In contrast, Generation Z designers with high NEU often exhibit caution and anxiety regarding
new technologies23, which decreases their positive perception of PEU and leads to lower WPS—a finding that
contradicts Liu et al. (2024)96. This contradiction may stem from differences in research contexts. In the study
by Liu et al. (2024)96, individuals with high levels of NEU may perceive online environments as more socially
inclusive due to reduced social pressure, thereby enhancing their PEU of technology. Furthermore, the effects
of CON and AGR on PEU were not significant. While this finding aligns with the research by Denden et al.
(2022)128, it differs from the conclusions of Agyei et al. (2020)71. This may be because designers, when using
AI tools, tend to focus more on creativity and functionality, making them less influenced by the impact of
conscientiousness or AGR on PEU. In contrast, in the mobile banking context studied by Agyei et al. (2020)71,
AGR and CON may encourage users to be more patient in learning and adapting to the technology, thereby
enhancing PEU. Finally, all traits except AGR and CON significantly impact PR. Specifically, OPE and EXT have
a substantial negative effect, while NEU positively affects PR. These findings are in line with some conclusions
from Whittingham et al. (2020)99 and Ahmed et al. (2021)94. Additionally, PR significantly mediates the
relationship between OPE, EXT, NEU, and WPS, leading Generation Z designers with higher levels of OPE
or EXT to be less sensitive to technological risks, subsequently increasing their WPS. However, the research
findings of Dangi & Singh (2021)64 indicate that OPE and EXT positively influence PR, contrasting with the
results of this study. This discrepancy may stem from the broader scope of online shopping risks examined in the
study, encompassing economic, health, and product quality concerns, while also lacking differentiation among
age groups. In traditional consumer contexts, individuals with higher levels of OPE and EXT tend to exhibit
greater sensitivity to uncertainty, thereby increasing their PR.
The SEM results indicate that, except for CON and AGR, the other three traits significantly influence WPS.
Specifically, OPE and EXT exhibit a significant positive impact on WPS, corroborating the findings of Maziriri
(2020)46 and Liu et al. (2024)96. This implies that Generation Z designers with elevated levels of these traits
are more inclined to subscribe to paid AI drawing tools. Conversely, NEU shows a significant negative impact
on WPS, suggesting that Generation Z designers with heightened levels of NEU tend to demonstrate lower
WPS. This conclusion is consistent with Qu et al. (2021)101 and Kumar et al. (2024)27, but contradicts Wang
et al. (2024)28. This difference may stem from the influence of cognitive load and emotional regulation on
the technology acceptance of NEU individuals. In the comparison between the utilitarian orientation of AI
drawing tools and the immersive experience of the VR learning environment in Wang et al. (2024)28, NEU
individuals exhibit varying levels of technology acceptance. CON and AGR did not exhibit significant effects on
WPS, conflicting with the conclusions of Kannen et al. (2024)45. This contradictory conclusion may stem from
differences in research subjects. Kannen et al. (2024)45 examined social media instant messaging services, which
emphasize planning and stability, whereas this study focuses on AI-based drawing tools. Users with high CON
may exhibit greater caution toward new technologies, leading to lower adoption intentions. Moreover, AGR, as a
socially oriented trait, influences technology acceptance based on whether users perceive the tool as facilitating
social interaction or collaboration. When a product (e.g., social media) enhances social value, AGR tends to
increase adoption intention. However, for tools primarily designed to enhance individual creative capabilities,
such as AI-based drawing tools, its influence may be weaker or even negligible. Furthermore, both PEU and
PU demonstrate significant positive impacts on WPS, indicating that ease of use, seamless functionality, and
practicality of paid AI drawing tools substantially enhance Generation Z designers’ WPS. This supports the
foundational assumptions of the TAM, underscoring the essential roles of PEU and PU in technology acceptance
and utilization2. PR shows a significant negative impact on WPS, aligning with extensive prior research6,66,
which implies that high-risk perception markedly reduces WPS. However, Jaradat et al. (2018)129 found that in
the adoption of smart government services, PR indirectly enhances usage intention through relative advantage,
trialability, and observability. Unlike AI drawing tools, citizens can more intuitively perceive the benefits of
smart government services, which alleviates concerns about security and privacy, making them more inclined
to use these services129.
The fsQCA analysis identified four combinations of conditions leading to WPS, highlighting key factors that
influence Generation Z designers’ willingness to subscribe to paid AI drawing tools under varied configurations.
Notably, configurations S1 and S2 each contain a single core condition that remains unaffected by the presence
or absence of other conditions. Configuration S1 displayed remarkably high original coverage and consistency,
further confirming the essential role of ~ FNEU as a necessary condition. Meanwhile, FOPE and FEXT were
identified as core conditions in configurations S3 and S4, respectively, emphasizing the significance of these
personality traits, in line with the SEM results. Additionally, the fsQCA results significantly diverge from the
SEM results in two main aspects. Firstly, FAGR plays a pivotal role as a core condition in configuration S2, a
stark contrast to the SEM results, where AGR is non-significant across the majority of relevant pathways. Its
emergence as a core condition in S2 suggests that AGR may serve as a prerequisite for other conditions, and
its absence might disrupt the overall mechanism, in alignment with traditional understandings84. Secondly,
although CON does not demonstrate a significant impact on WPS in the SEM results, it appears as an edge
condition in two of the four pathways identified by fsQCA, indicating that SEM does not capture the complex
combination effects of CON with other personality traits. In summary, the fsQCA results offer richer and more
nuanced insights, complementing the findings of SEM and highlighting the value of using multiple methods to
analyze data for a more comprehensive understanding of complex variable relationships.

Scientific Reports | (2025) 15:17587 | [Link] 13

[Link]/scientificreports/

Theoretical implications
The theoretical contributions of this study are reflected in several key aspects. Firstly, it significantly expands the
existing body of literature by examining the impact of BFP traits on the willingness of Generation Z designers to
subscribe to paid AI drawing tools, thereby deepening and broadening research in personality psychology and
technology acceptance. Previous studies on the relationship between personality traits and technology adoption
have primarily focused on fields such as green banking60,71, e-commerce24,25, and social media45,46. However,
research specifically addressing AI technologies remains relatively scarce. Most existing research on AI tools
has focused on market potential5,29 and user acceptance2,6,30, but the exploration of users’ WPS is still nascent.
Systematically, this study reveals, for the first time, how the personality traits of Generation Z designers—a
specific demographic group—influence their adoption behavior concerning paid AI drawing tools, thereby
addressing a significant research gap.
Secondly, this study proposes a comprehensive theoretical framework by integrating the BFP model, the
TAM, and the PRT. It develops an extended model to explore the differential mechanisms of various personality
traits on PEU, PU, and PR, and their direct impact on WPS. Furthermore, this study validates key hypotheses
and related findings of TAM, providing substantial theoretical support for classification-based research on
personality traits in the context of emerging technology adoption. Additionally, it addresses the limited research
involving PEU, PU, and PR as mediating variables between BFP and WPS. By introducing these perceptual
factors, an exploratory analysis was conducted to determine their mediating role between individual traits and
behavioral decision-making. This yields valuable empirical evidence for future studies on how individual traits
influence behavioral decision-making through perceptual factors, enriching the theoretical understanding of
these processes.
Finally, this study is innovative in its methodology. Building on the principles of complexity theory, which
emphasizes the nonlinear interaction of multiple dynamic factors within systems79, this study effectively
addresses the complex phenomenon of multiple concurrent causal relationships using fsQCA. By employing
fsQCA, the study reveals the multi-path causal relationships between different BFP configurations and the
WPS among Generation Z designers, while also comparing these findings with the results of SEM. This hybrid
research approach not only overcomes the limitations of traditional Multi-Regression Analysis (MRA) and SEM
but also serves as a valuable reference for future studies in related fields. Moreover, the contrasting performances
of CON and AGR in SEM and fsQCA underscore the need for further research into the complex effects of
these two personality traits on user adoption of AI technologies across different contexts and demographic
groups. Therefore, this study not only provides an in-depth understanding of the mechanisms through which
personality traits influence Generation Z designers’ willingness to subscribe to paid AI drawing tools but also
contributes significantly to the methodology of AI technology acceptance and user behavior research.

Managerial implications
From a managerial perspective, this study offers several crucial insights. Firstly, Generation Z, as the demographic
with the highest application rate of AI technology17, is a key representative for this research. The SEM results
demonstrate the varied impacts of BFP traits on PU, PEU, PR, and WPS, suggesting that developers should
segment the market based on personality traits60. For instance, users with high levels of OPE and EXT are more
likely to find paid AI drawing tools user-friendly and low-risk, and therefore are more inclined to subscribe.
These users should be targeted in marketing strategies. For users with high levels of NEU, developers can mitigate
their concerns by providing clear user guides, awareness campaigns, and training programs that highlight the
advantages of the tools71. Additionally, the study indicates that PEU and PU positively affect WPS, whereas PR
negatively influences WPS. These insights provide direction for optimizing product design and strategies for
managing user trust in paid AI drawing tools. Developers should focus on feature completeness and ease of
operation during the design phase, while also enhancing transparent information sharing and risk management
practices. These measures can reduce users’ perception of risk6, build their trust in the product, and ultimately
increase their willingness to pay.
The fsQCA results presented herein reveal four primary paths to achieving high WPS among Generation
Z designers, underscoring the inadequacy of a ‘one size fits all’ strategy113. Developers can enhance user
segmentation using the identified configurations. Specifically, special attention and tailored marketing strategies
should be directed toward users with lower levels of NEU. Although AGR did not significantly impact WPS
in the SEM analysis, fsQCA recognizes it as a pivotal condition, suggesting that users exhibiting higher AGR
in certain trait combinations should also be prioritized. For users displaying high levels of OPE and CON, or
CON and EXT, these configurations, while showing limited coverage in fsQCA, represent potential markets.
Developers should leverage refined marketing strategies to foster subscription intentions in these segments,
ultimately boosting market penetration and user satisfaction.
Based on these insights, it is recommended that developers of paid AI drawing tools offer users a free trial
period and employ strategies akin to those used by MidJourney and Leonardo AI, which facilitate access and
operation through social platforms such as Discord. During this trial period, developers are encouraged to
utilize advanced analytical methods, including Chaurasia et al.‘s (2024)130 multimodal attention-based CNN-
BiLSTM model and Wang et al.‘s (2025)131 unimodal emotion-temporal long short-term memory (LSTM)
network, to process text and video data from user interactions on social platforms. These techniques allow for
precise identification and segmentation of user traits, which aids in crafting more personalized user engagement
strategies.
Finally, this study offers practical guidance for education and human resource management sectors. Within
education, colleges and institutions can customize their teaching strategies to align with the personality traits of
design students28, addressing variations in technology acceptance and usage preferences among students with
differing traits. For example, students with high levels of EXT and OPE might benefit from an emphasis on

Scientific Reports | (2025) 15:17587 | [Link] 14

[Link]/scientificreports/

applying AI tools in creative design and enhancing efficiency through project-based learning, which encourages
initiative and innovation. Conversely, for students with higher NEU levels, educational strategies should focus
on safety and compliance to reduce their technology-related anxiety63. In terms of recruitment and team
management, the study suggests using a personality-trait-based approach to screen potential AI tool users.
Enterprises can evaluate personality traits to identify designers suited for AI tool adoption, fostering them as
key innovators and speeding up the organization’s AI adoption. Additionally, companies can devise technology
support strategies tailored to various personality traits, optimizing team technology use and bolstering long-
term organizational innovation.

Limitations and future research

This study has several limitations that suggest directions for future research. First, the sample is limited to
Generation Z designers, restricting external validity. Future studies could extend the scope to different age
groups and industry sectors to enhance generalizability. Second, as the data primarily relies on self-reported
measures, potential social desirability bias may exist. Future research could integrate experimental methods,
longitudinal tracking, or big data analytics to improve objectivity and robustness. Additionally, this study
employs cross-sectional data, which fails to capture the dynamic changes in user behavior. Future research could
adopt longitudinal designs or machine learning approaches, such as time series analysis and cluster analysis,
to model the evolution of subscription behavior, uncover long-term adoption mechanisms, and enhance
predictive accuracy. Furthermore, this study does not fully account for the privacy and ethical risks associated
with AI drawing tools, including data security, algorithmic bias, and copyright concerns. Future research could
further explore their impact on user trust and long-term engagement. Lastly, this study does not incorporate
potential moderating factors such as gender, cultural background, financial constraints, subscription experience,
educational background, brand loyalty, tool functionality, and pricing diversity, all of which may significantly
influence user decision-making. Future research could adopt multilevel analysis to optimize variable selection
and model construction, thereby offering a more comprehensive theoretical perspective and practical insights.

Data availability
The datasets used and/or analyzed during the current study available from the corresponding author on reason-
able request. ([Link]

Received: 11 January 2025; Accepted: 16 May 2025

References
1. Wingström, R., Hautala, J. & Lundman, R. Redefining creativity in the era of AI? Perspectives of computer scientists and new
media artists. Creativity Res. J. 36 (2), 177–193 (2024).
2. 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), e23305 (2024).
3. Chesher, C. & Albarrán-Torres, C. The emergence of autolography: the ‘magical’invocation of images from text through AI.
Media Int. Aust. 189 (1), 57–73 (2023).
4. Wang, Y., Zhao, Y., Tian, X., Yang, J. & Luo, S. The influence of subjective knowledge, technophobia and perceived enjoyment on
design students’ intention to use artificial intelligence design tools. Int J. Technol. Des. Ed. 35, 333–358 (2024).
5. 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).
6. 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 Int. 41(2), 1064–1077 (2024).
7. Lesmana, V. A. A., Tina, A. & Retno Yanti, S. Optimizing AI’s role in advancing interior design industry. J. Artif. Intell. Archit. 3
(2), 61–71 (2024).
8. Homeyer, A. et al. Artificial intelligence in pathology: from prototype to product. J. Pathol. Inf. 12 (1), 13 (2021).
9. Wang, Y. & Zhang, W. Factors influencing the adoption of generative AI for Art designing among Chinese generation Z: A
structural equation modeling approach. Ieee Access. 11, 143272–143284 (2023).
10. Wang, S. F. & Chen, C. C. Exploring designer trust in artificial Intelligence-Generated content: TAM/TPB model study. Appl. Sci.
14 (16), 6902 (2024).
11. 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. Inf. Syst. J. 30 (2), 295–333 (2020).
12. Campbell, W. K., Campbell, S. M., Siedor, L. E. & Twenge, J. M. Generational differences are real and useful. Ind. Organ. Psychol.
8 (3), 324–331 (2015).
13. Maloni, M., Hiatt, M. S. & Campbell, S. Understanding the work values of gen Z business students. Int. J. Manag Educ. 17 (3),
100320 (2019).
14. Herring, H. Connecting Generations: Bridging the Boomer, Gen X, and Millennial Divide (Rowman & Littlefield, 2019).
15. Hernandez-de-Menendez, M. & Díaz, E. Morales-Menendez, R. Educational experiences with generation Z. Int. J. Interact. Des.
Manuf. 14 (3), 847–859 (2020).
16. Udemy Business. Gen Z Enters the Workforce: Welcoming the New Generation of Workplace Elites. (2024). Retrieved from: ​h​t​tp ​ ​s​
:​​/​/​b​u​s​in
​​ ​e​s​s​.​u​​d​e​m​y​.​c​​o​m​/​z​h​​-​h​a​n​s​/​​r​e​s​o​ur​ ​​c​e​s​/​g​e​​n​-z​ ​-​i​​n​-​t​h​e​-w ​​ ​o​r​k​p​l​​a​c​e​-​r​e​​p​o​r​t​/​​?​u​t​m​_​m​​e​d​i​u​m=
​ ​​d​i​r​e​c​t​​&​ut​ ​m​_​s​o​u​r​c​e​=​d​i​r​e​c​t Accessed
31 Aug 2024.
17. Microsoft & LinkedIn. Work Trend Index Annual Report. (2024). Retrieved from: ​h​tt​ ​p​s​:​​​/​​/​w​ww ​​ ​.​m​i​c​r​o​s​o​f​​t ​.​c​​​o​m​/​​e​​n​​-​u​s​/​w​​o​r​kl​​ ​a​​b​/​​w​
o​r​​k​-​t​​r​e​​n​d​-​​i​nd ​ ​​​e​x​​/​a​i​​-​a​t​-​​​w​o​r​k​-​​is​ ​-​​h​e​r​e​-​n​​ow
​ ​-​​c​o​m​e​s​​-​th
​ ​e​-​h​a​r​d​-​p​a​r​t (2024). Accessed 2 Sep 2024.
18. Ho, M. T. et al. Rethinking technological acceptance in the age of emotional AI: surveying gen Z (Zoomer) attitudes toward non-
conscious data collection. Technol. Soc. 70, 102011 (2022).
19. Chan, C. K. Y. & Lee, K. K. W. The AI generation Gap: are gen Z students more interested in adopting generative AI such as
ChatGPT in teaching and learning than their gen X and millennial generation teachers? Smart Learn. Environ. 10 (1), 60 (2023).
20. Magano, J. et al. Generation Z: fitting project management soft skills Competencies—A Mixed-Method approach. Educ. Sci. 10
(7), 187 (2020).
21. Saha, P., Sengupta, A. & Gupta, P. Influence of personality traits on generation Z consumers’ click-through intentions towards
personalized advertisements: A mixed-methods study. Heliyon 10(15), e34559 (2024).

Scientific Reports | (2025) 15:17587 | [Link] 15

[Link]/scientificreports/

22. Roberts, B. W. & Yoon, H. J. Personality Psychol. Annu Rev. Psychol 73, 489–516 (2022).
23. Lissitsa, S. & Kol, O. Four generational cohorts and hedonic m-shopping: association between personality traits and purchase
intention. Electron. Commer. Res. 21 (2), 545–570 (2021).
24. Burhan, N. A. S. et al. Materialism and Big-Five personality traits shaping low-income university students’ compulsive online-
buying behavior. Pertanika J. Social Sci. Humanit. 30 (4), 1931–1953 (2022).
25. Bashir, T., Zhongfu, T., Sadiq, B., Anwar, A. & Naseem, A. Personality traits and its impact on continuance intention to use social
networking sites to buy branded clothing. Ind. Textila. 74 (5), 564–571 (2023).
26. Srivastava, A., Dasgupta, S. A., Ray, A., Bala, P. K. & Chakraborty, S. Relationships between the big five personality types and
consumer attitudes in Indian students toward augmented reality advertising. Aslib J. Inf. Manag. 73 (6), 967–991 (2021).
27. Kumar, A., Shankar, A. & Nayal, P. Metaverse is not my cup of tea! An investigation into how personality traits shape metaverse
usage intentions. J. Retail Consum. Serv. 77, 103639 (2024).
28. Wang, Y. M. et al. Determinants of students’ adoption of virtual reality-based learning systems: an individual difference
perspective. Innov. Educ. Teach. Int. 61 (1), 143–153 (2024).
29. Ali, O., Murray, P. A., Momin, M., Dwivedi, Y. K. & Malik, T. The effects of artificial intelligence applications in educational
settings: challenges and strategies. Technol. Forecast. Soc. 199, 123076 (2024).
30. 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).
31. Zheng, W., Qiu, H. & Morrison, A. M. Applying a combination of SEM and FsQCA to predict tourist Resource-Saving behavioral
intentions in rural tourism: an extension of the theory of planned behavior. Int. J. Env Res. Pub He. 20 (2), 1349 (2023).
32. Pappas, I. O. & Woodside, A. G. Fuzzy-set qualitative comparative analysis (fsQCA): guidelines for research practice in
information systems and marketing. Int. J. Inf. Manage. 58, 102310 (2021).
33. Tavitiyaman, P., Zhang, X. & Tsang, W. Y. How tourists perceive the usefulness of technology adoption in hotels: interaction effect
of past experience and education level. J. China Tour Res. 18 (1), 64–87 (2022).
34. Kim, Y. W., Lim, C. & Ji, Y. G. Exploring the user acceptance of urban air mobility: extending the technology acceptance model
with trust and service quality factors. Int. J. Hum. -Comput Int. 39 (14), 2893–2904 (2023).
35. Davis, F. D. & Perceived Usefulness Perceived ease of use, and user acceptance of information technology. Mis Quart. 13 (3),
319–340 (1989).
36. Fishbein, M. & Ajzen, I. Belief, Attitude, Intention and Behaviour: an Introduction To Theory and Research (Addison-Wesley
Publishing, 1975).
37. Li, G., Liang, Y., Wang, H., Chen, J. & Chang, X. Factors Influencing Users’ Willingness to Adopt Connected and Autonomous
Vehicles: Net and Configurational Effects Analysis Using PLS-SEM and FsQCA. J. Adv. Transport. 7489897 (2022). (2022)(1).
38. Basarir-Ozel, B., Nasir, V. A. & Turker, H. B. Determinants of smart home adoption and differences across technology readiness
segments. Technol. Forecast. Soc. 197, 122924 (2023).
39. Santiago, J., Borges-Tiago, M. T. & Tiago, F. Embracing RAISA in restaurants: exploring customer attitudes toward robot adoption.
Technol. Forecast. Soc. 199, 123047 (2024).
40. Kumpulainen, H., Kinnunen, U. M. & Jylhä, V. ICT acceptance in social and healthcare context–A scoping review of theories and
models. Int. J. Med. Inf. 195, 105796 (2025).
41. Symasek, L., Yeazitzis, T., Weger, K. & Mesmer, B. Recent developments in individual difference research to inform the adoption
of AI technology. Systems 13 (3), 156 (2025).
42. Chen, S. & Ye, J. Understanding consumers’ intentions to purchase smart clothing using PLS-SEM and FsQCA. Plos One. 18 (9),
e0291870 (2023).
43. Al-Okaily, M. ChatGPT as an educational resource for accounting students: expanding the classical TAM model. Educ Inf.
Technol. [Link] (2025).
44. Holdack, E., Lurie-Stoyanov, K. & Fromme, H. F. The role of perceived enjoyment and perceived informativeness in assessing the
acceptance of AR wearables. J. Retail Consum. Serv. 65, 102259 (2022).
45. Kannen, C., Sindermann, C. & Montag, C. On the willingness to pay for social media/messenger services taking into account
personality and sent/received messages among whatsapp users. Heliyon 10(9) (2024).
46. Maziriri, E. T. The big five personality traits influencing habitual Facebook usage, life satisfaction and psychological well-being of
generation Y students. Td-J Transdiscipl Re. 16 (1), 1–13 (2020).
47. Minssen, L., Levels, M., Pfeifer, H. & Wehner, C. Recruiting mid-qualified workers in product-innovating firms: which personality
traits matter? J. Behav. Exp. Econ. 112, 102267 (2024).
48. Hickman, L. et al. Developing and evaluating language-based machine learning algorithms for inferring applicant personality in
video interviews. Hum. Resour. Manag J. 34 (2), 255–274 (2024).
49. Maran, T. K. et al. Who fits into the digital workplace? Mapping digital self-efficacy and agility onto psychological traits. Technol.
Forecast. Soc. 175, 121352 (2022).
50. Kottirre, J., Blickle, G. & Wihler, A. Fearless happiness in apparel store sales performance and the two faces of conscientiousness.
Pers. Indiv Differ. 232, 112864 (2025).
51. Shahzad, K., Raja, U. & Hashmi, S. D. Impact of big five personality traits on authentic leadership. Leadersh. Org. Dev. J. 42 (2),
208–218 (2021).
52. Fosse, T. H. et al. Neuroticism as an antecedent of abusive supervision and laissez-faire leadership in emergent leaders: the role of
facets and agreeableness as a moderator. Appl. Psychol. 73 (2), 675–697 (2024).
53. Duong, C. D. Big five personality traits and green consumption: bridging the attitude-intention-behavior gap. Asia Pac. J. Market
Lo. 34 (6), 1123–1144 (2022).
54. Shetty, A. & Rizwana, M. Embracing sustainability on Indian roads: exploring the influence of personality factors in fostering
green transportation behaviour. Int. J. Sustain. Energy. 43 (1), 2365340 (2024).
55. Meyer, J., Jansen, T., Hübner, N. & Lüdtke, O. Disentangling the association between the big five personality traits and student
achievement: Meta-Analytic evidence on the role of domain specificity and achievement measures. Educ. Psychol. Rev. 35 (1), 12
(2023).
56. Shaninah, F. S. E. & Mohd Noor, M. H. The impact of big five personality trait in predicting student academic performance. J.
Appl. Res. High. Educ. 16 (2), 523–539 (2024).
57. Sultan, S. et al. Relationship of big five personality traits and future specialty preference among undergraduate medical students:
a cross-sectional study. Egypt. J. Neurol. Psych. 59 (1), 103 (2023).
58. Al Sawafi, A., Al Lawati, A. & Al Sinawi, H. Personality traits and potential career choices among medical students at Sultan
Qaboos University: A Cross-Sectional study. Cureus 16(1), e51994 (2024).
59. Watjatrakul, B. Intention to adopt online learning. Int. J. Inf. Learn. Tech. 37 (1/2), 46–65 (2020).
60. Malik, G. & Singh, D. Personality matters: does an individual’s personality affect adoption and continued use of green banking
channels? Int. J. Bank. Mark. 40 (4), 746–772 (2022).
61. Bauer, R. A. Consumer behavior as risk taking. Marketing: Crit. Perspect. Bus. Manage. 593, 13–21 (1967).
62. Almaiah, M. A. et al. Investigating the role of perceived risk, perceived security and perceived trust on smart m-Banking
application using SEM. Sustainability 15 (13), 9908 (2023).
63. Changchit, C., Cutshall, R., DeLatte, G., Ibrahimov, A. & Changchit, C. Exploring personality traits’ impact on risk perception in
mobile payments. J Comput. Inf. Syst. [Link] (2024).

Scientific Reports | (2025) 15:17587 | [Link] 16

[Link]/scientificreports/

64. Dangi, A. & Singh, V. Personality influences risk perception in online shopping: an Indian consumer perspective. Int. J. E-Bus Res.
17 (3), 1–12 (2021).
65. Al Adwan, A. S. et al. Extending the technology acceptance model (TAM) to predict university students’ intentions to use
Metaverse-Based learning platforms. Educ. Inf. Technol. 28 (11), 15381–15413 (2023).
66. Bu, F., Wang, N., Jiang, Q. & Tian, X. Research on Privacy-by-Design behavioural Decision-Making of information engineers
considering perceived work risk. Systems 12 (7), 250 (2024).
67. Brougham, D. & Haar, J. Smart technology, artificial intelligence, robotics, and algorithms (STARA): employees’ perceptions of
our future workplace. J. Manage. Organ. 24 (2), 239–257 (2018).
68. Kong, H. et al. Influences of artificial intelligence (AI) awareness on career competency and job burnout. Int. J. Contemp. Hosp.
M. 33 (2), 717–734 (2021).
69. Mo, Z., Liu, M. T. & Ma, Y. How AI awareness can prompt service performance adaptivity and technologically-environmental
mastery. Tourism Manage. 105, 104971 (2024).
70. Liu, S. & Cheng, P. The double-edged sword effect of artificial intelligence awareness among hotel employees. Int. J. Contemp.
Hosp. M. 37 (3), 997–1015 (2025).
71. Agyei, J., Sun, S., Abrokwah, E., Penney, E. K. & Ofori-Boafo, R. Mobile banking adoption: examining the role of personality
traits. Sage Open. 10 (2), 2158244020932918 (2020).
72. Pavlíček, A., Bobenič Hintošová, A. & Sudzina, F. Impact of personality traits and demographic factors on risk attitude. Sage
Open. 11 (4), 21582440211066917 (2021).
73. McSharry, P. & Mawejje, J. Estimating urban GDP growth using nighttime lights and machine learning techniques in data poor
environments: the case of South Sudan. Technol. Forecast. Soc. 203, 123399 (2024).
74. Zyphur, M. J., Bonner, C. V. & Tay, L. Structural equation modeling in organizational research: the state of our science and some
proposals for its future. Annu. Rev. Organ. Psych. 10, 495–517 (2023).
75. Goldstein, J. Emergence as a construct: history and issues. Emergence 1 (1), 49–72 (1999).
76. Goldstein, J., Hazy, J. & Lichtenstein, B. Complexity and the Nexus of Leadership: Leveraging Nonlinear Science To Create Ecologies
of Innovation (Springer, 2010).
77. Turner, J. R. & Baker, R. M. Complexity theory: an overview with potential applications for the social sciences. Systems 7 (1), 4
(2019).
78. Schneider, M. & Somers, M. Organizations as complex adaptive systems: implications of complexity theory for leadership
research. Leadersh. Quart. 17 (4), 351–365 (2006).
79. Castellani, B. & Hafferty, F. W. Sociology and Complexity Science: A New Field of Inquiry (Springer Science & Business Media,
2009).
80. Martín-Rojas, R., Garrido-Moreno, A. & García-Morales, V. J. Social media use, corporate entrepreneurship and organizational
resilience: A recipe for SMEs success in a post-Covid scenario. Technol. Forecast. Soc. 190, 122421 (2023).
81. Wang, F., Jiang, J. & Cosenz, F. Understanding data-driven business model innovation in complexity: A system dynamics
approach. J. Bus. Res. 186, 114967 (2025).
82. Tjörnbo, O. & McGowan, K. A complex-systems perspective on the role of universities in social innovation. Technol. Forecast. Soc.
174, 121247 (2022).
83. Vasist, P. N. & Krishnan, S. Powered by innovation, derailed by disinformation: A multi-country analysis of the influence of
online political disinformation on nations’ innovation performance. Technol. Forecast. Soc. 199, 123029 (2024).
84. Qu, Y., Wang, W. & Han, R. A multi-method study of the emotional mechanism linking seaside destination attributes and tourists’
revisit intention. Curr. Issues Tour. 27 (11), 1834–1851 (2024).
85. De Canio, F., Viglia, G. & Lombart, C. Package-free products: how to improve pro-environmental buying behaviors among
consumers. Technol. Forecast. Soc. 198, 122995 (2024).
86. Gilead, T. Conceptualizing distributive justice in education: a complexity theory perspective. J. Philos. Educ. 57 (2), 495–516
(2023).
87. Wang, Q. C. et al. The impact of personality traits on household energy conservation behavioral intentions–An empirical study
based on theory of planned behavior in Xi’an. Sustain. Energy Techn. 43, 100949 (2021).
88. George, J. M. & Zhou, J. When openness to experience and conscientiousness are related to creative behavior: an interactional
approach. J. Appl. Psychol. 86 (3), 513 (2001).
89. Chandra, B. Technology acceptance and self-enhancement in social media. Multimed Tools Appl. 83 (31), 75483–75509 (2024).
90. Lee, C. K. & Foo, M. Y. Risk attitudes and the big five personality traits: a study on construction project practitioners in Malaysia.
Int. J. Constr. Manag. 22 (13), 2435–2445 (2022).
91. Wu, R. & Yu, Z. The influence of social isolation, technostress, and personality on the acceptance of online meeting platforms
during the COVID-19 pandemic. Int. J. Hum. -Comput Int. 39 (17), 3388–3405 (2023).
92. Alsyouf, A., Ishak, A. K., Lutfi, A., Alhazmi, F. N. & Al-Okaily, M. The role of personality and top management support in
continuance intention to use electronic health record systems among nurses. Int. J. Env Res. Pub He. 19 (17), 11125 (2022).
93. Moslehpour, M., Pham, V. K., Wong, W. K. & Bilgiçli, İ. e-Purchase Intention of Taiwanese Consumers: Sustainable Mediation of
Perceived Usefulness and Perceived Ease of Use. Sustainability, 10(1), 234 (2018).
94. Ahmed, S., Ramzan, M., Sheikh, A. & Ali, A. Impact of students personality traits on social networking sites usage, benefits and
risks. Glob Knowl. Mem. Commu. 70 (6/7), 518–537 (2021).
95. Irfan, M. & Ahmad, M. Modeling consumers’ information acquisition and 5G technology utilization: is personality relevant? Pers.
Indiv Differ. 188, 111450 (2022).
96. Liu, A. Y. & Carrubi, B. Blanco González-Tejero, C. Community startup businesses: the impact of big five personality traits and
social media technology acceptance on group buying leaders. Int. J. Entrep Behav. R. 30 (4), 955–978 (2024).
97. Wang, C. M., Xu, B. B., Zhang, S. J. & Chen, Y. Q. Influence of personality and risk propensity on risk perception of Chinese
construction project managers. Int. J. Proj Manag. 34 (7), 1294–1304 (2016).
98. Gosling, S. D., Rentfrow, P. J. & Swann, W. B. A very brief measure of the Big-Five personality domains. J. Res. Pers. 37 (6),
504–528 (2003).
99. Whittingham, N., Boecker, A. & Grygorczyk, A. Personality traits, basic individual values and GMO risk perception of Twitter
users. J. Risk Res. 23 (4), 522–540 (2020).
100. Jeronimus, B. F., Riese, H., Sanderman, R. & Ormel, J. Mutual reinforcement between neuroticism and life experiences: a five-
wave, 16-year study to test reciprocal causation. J. Pers. Soc. Psychol. 107 (4), 751–764 (2014).
101. Qu, W., Sun, H. & Ge, Y. The effects of trait anxiety and the big five personality traits on self-driving car acceptance. Transportation
48 (5), 2663–2679 (2021).
102. Azzahra, T. R. & Kusumawati, N. The impact of mobile service quality, perceived value, perceived usefulness, perceived ease of
use, customer satisfaction towards continuance intention to use MyTelkomsel app. J. Consumer Stud. Appl. Mark. 1 (1), 46–60
(2023).
103. Qin, X., Shi, Y., Lyu, K. & Mo, Y. Using a TAM-TOE model to explore factors of Building information modelling (BIM) adoption
in the construction industry. J. Civ. Eng. Manag. 26 (3), 259–277 (2020).
104. Lei, P. W. & Wu, Q. Introduction to structural equation modeling: issues and practical considerations. Educ. Meas-Issues Pra. 26
(3), 33–43 (2007).

Scientific Reports | (2025) 15:17587 | [Link] 17

[Link]/scientificreports/

105. Henseler, J. & Schuberth, F. Should PLS become factor-based or should CB-SEM become composite-based? Both! Eur J. Inf. Syst.
[Link] (2024).
106. 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).
107. 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).
108. Shamout, M. D. Supply chain data analytics and supply chain agility: a fuzzy sets (fsQCA) approach. Int. J. Organ. Anal. 28 (5),
1055–1067 (2020).
109. Misangyi, V. F. et al. Embracing causal complexity: the emergence of a neo-configurational perspective. J. Manage. 43 (1), 255–
282 (2017).
110. Kraus, S., Ribeiro-Soriano, D. & Schüssler, M. Fuzzy-set qualitative comparative analysis (fsQCA) in entrepreneurship and
innovation research – the rise of a method. Int. Entrep Manag J. 14 (1), 15–33 (2018).
111. Olan, F., Liu, S., Neaga, I., Chen, H. & Nakpodia, F. How cultural impact on knowledge sharing contributes to organizational
performance: using the FsQCA approach. J. Bus. Res. 94, 313–319 (2019).
112. Um, T., Chung, N. & Stienmetz, J. Factors affecting consumers’ impulsive buying behavior in tourism mobile commerce using
SEM and FsQCA. J. Vacat Mark. 29 (2), 256–274 (2023).
113. Liu, X., Zhang, C. & Wu, J. Explaining consumers’ continuous purchase intention toward subscriber-based knowledge payment
platforms: findings from PLS-SEM and FsQCA. Aslib J. Inf. Manag. 76 (2), 189–211 (2024).
114. Xie, X. Z. & Tsai, N. C. The effects of negative information-related incidents on social media discontinuance intention: evidence
from SEM and FsQCA. Telemat Inf. 56, 101503 (2021).
115. Brecher, C. et al. Nachhaltige Produktion-as-a-Service. In Aachener Werkzeugmaschinen-Kolloquium 2023 (2023).
116. Sekaran, U. & Bougie, R. Research Methods for Business: A Skill Building Approach (Wiley, 2016).
117. Hair, J. F., Black, W. C. & Babin, B. J. Multivariate Data Analysis: A Global Perspective 7th edn (Pearson Education, 2010).
118. Faul, F., Erdfelder, E., Buchner, A. & Lang, A. G. Statistical power analyses using G*Power 3.1: tests for correlation and regression
analyses. Behav. Res. Methods. 41 (4), 1149–1160 (2009).
119. Heine, S. J., Buchtel, E. E. & Personality The universal and the culturally specific. Annu. Rev. Psychol. 60, 369–394 (2009).
120. 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).
121. Podsakoff, P. M. & Organ, D. W. Self-reports in organizational research: problems and prospects. J. Manage. 12 (4), 531–544
(1986).
122. Spiliotopoulou, G. Reliability reconsidered: Cronbach’s alpha and paediatric assessment in occupational therapy. Aust Occup.
Ther. J. 56 (3), 150–155 (2009).
123. Cole, D. A. Utility of confirmatory factor analysis in test validation research. J. Consult Clin. Psych. 55 (4), 584 (1987).
124. Fornell, C. & Larcker, D. F. Evaluating structural equation models with unobservable variables and measurement error. J. Mark.
Res. 18 (1), 39–50 (1981).
125. Hayes, A. F. Beyond Baron and Kenny: statistical mediation analysis in the new millennium. Commun. Monogr. 76 (4), 408–420
(2009).
126. Ragin, C. C. Redesigning Social Inquiry: Fuzzy Sets and Beyond (University of Chicago Press, 2008).
127. Fiss, P. C. Building better causal theories: A fuzzy set approach to typologies in organization research. Acad. Manage. J. 54 (2),
393–420 (2011).
128. Denden, M. et al. To use or not to use: impact of personality on the intention of using gamified learning environments. Electronics
11 (12), 1907 (2022).
129. Jaradat, M. I. R. M., Moustafa, A. A. & Al-Mashaqba, A. M. Exploring perceived risk, perceived trust, perceived quality and the
innovative characteristics in the adoption of smart government services in Jordan. Int. J. Mob. Commun. 16 (4), 399–439 (2018).
130. Chaurasia, S., Bharti, K. K. & Gupta, A. A multi-model attention based CNN-BiLSTM model for personality traits prediction
based on user behavior on social media. Knowl. -Based Syst. 300, 112252 (2024).
131. Wang, J., Li, H., Woo, W. L. & Shan, S. A single modality apparent first impression personality recognition model with Temporal
emotion based LSTM. Expert Syst. Appl. 259, 125114 (2025).

Author contributions
N.D.: Conceptualization, Formal analysis, Investigation, Writing – original draft, Writing – review & editing.
M.W.C.: Methodology, Resources, Supervision, Writing – review & editing. L.L.H.: Data curation, Conceptual-
ization, Validation, Visualization. All authors reviewed the manuscript.

Declarations

Competing interests
The authors declare no competing interests.

Additional information
Supplementary Information The online version contains supplementary material available at ​ht​ ​t​p​s​:​/​/​d​o​i.​ ​o​r​g​/​1​
0​.​1​03​ ​8​/​s​4​1​5​9​8​-0​ ​2​5​-​0​2​9​2​6​-​8​​​​.​​
Correspondence and requests for materials should be addressed to M.C.
Reprints and permissions information is available at [Link]/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Scientific Reports | (2025) 15:17587 | [Link] 18

[Link]/scientificreports/

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 ​h​t​t​p​:​//​ ​c​r​e​a​t​i​v​e​co
​ ​m​m​o​
n​s​.​o​r​g/​ ​l​i​c​e​n​s​e​s​/b
​ ​y​-​n​c​-​n​d​/​4​.​0​/​​​​.​​

© The Author(s) 2025

Scientific Reports | (2025) 15:17587 | [Link] 19