Exploring Chinese Millennials’ Purchase Intentions for Clothing with AI-Generated Patterns from Premium Fashion Brands: An Integration of the Theory of Planned Behavior and Perceived Value Perspective

Abstract

Premium fashion brands are increasingly adopting Generative Artificial Intelligence (GenAI) to reduce costs and enhance creativity. However, consumers have mixed perceptions of clothing with AI-generated patterns (CAGPs) launched by premium fashion brands, especially in online shopping contexts where consumers cannot examine physical products firsthand. This study integrates the Theory of Planned Behavior (TPB) with Customer Perceived Value (CPV) to investigate Chinese Millennials’ attitudes and purchase intentions toward online purchases of CAGPs launched by premium fashion brands. Using a purposive sampling approach, the study collected 471 valid responses from Chinese Millennials. Structural equation modeling (SEM) was then employed to test the proposed model and hypotheses. The results reveal that perceived brand design effort and perceived price value are primary drivers of purchase intention for CAGPs from premium fashion brands, while perceived aesthetic value significantly shapes consumer attitudes. The subjective norm and attitude positively influence purchase intention. This study sheds light on the roles of aesthetic, emotional, monetary and social factors in driving purchase intention, offering practical suggestions for premium brands’ product design and marketing strategies.

1. Introduction

In the era of intelligent digitalization, Generative Artificial Intelligence (GenAI) has become a transformative force in the fashion design industry [1]. GenAI models generate creative textile and fabric patterns from input data (e.g., text or images) using natural language processing and deep learning techniques [2]. In visually driven online shopping environments, the use of GenAI-generated patterns not only expands creative possibilities and enhances design efficiency but also reduces labor costs [1]. As a result, it has become increasingly favored by clothing brands that prioritize online sales channels.

Despite its growing adoption, the application of GenAI in designing patterns for premium clothing presents several challenges. Premium fashion brands typically emphasize brand values such as originality, cultural significance, and artisanal craftsmanship. However, the “non-human” nature of GenAI-generated content may be perceived as undermining authenticity [3], thereby triggering consumer concerns regarding brand ethics [4], product quality [3], and brand image [5]. For example, when Selkie released its GenAI-generated online collection The Golden Paradise in 2024, one particular item—a sweater featuring a six-toed puppy—sparked public backlash. Consumers criticized the design for breaching conventional aesthetic norms, raising concerns about both its appropriateness and pricing. This phenomenon reflects consumers’ sensitivity to both the extent of AI involvement in pattern design and the perceived rationality of its outcomes. It also suggests that although GenAI offers creative breakthroughs for premium fashion brands, its market acceptance ultimately depends on consumers’ trust in the quality of AI-generated creations and is evaluated based on the perceived value they deliver.

Consumer concerns about brand originality and perceived apparel value—driven by AI-dominated pattern design—are growing. This trend may be particularly evident in China, the world’s largest fashion consumer market, where Millennial consumers hold a dominant position, accounting for 61.8% of total clothing purchases [6]. This demographic (born between 1981 and 1996) shows a strong preference for online shopping [7]. Compared to other consumer groups, Chinese Millennials are more interested in technological innovation, personalized design, and brand experience [8]. They also tend to be early adopters of AI-generated fashion products [9]. Although they demonstrate openness toward AI and exhibit brand loyalty to premium labels, it remains unclear whether Chinese Millennials continue to hold positive attitudes and maintain purchase intentions when premium fashion brands delegate part of their pattern design process to GenAI technologies. This issue has yet to be systematically explored. This not only concerns clarifying the mechanisms underlying consumer attitudes in the context of AI but also influences how premium fashion brands construct new human–AI collaborative design pathways and redefine the focal points of value communication.

Although research on consumer responses to AI-generated products is expanding [9,10,11], most existing studies still focus on applications such as AI-powered fashion recommendation services [12], personalized customization [13], and virtual try-on technologies [14]. Relatively little attention has been paid to the use of AI in premium fashion products. Moreover, previous investigations have focused mainly on isolated aspects of consumer value perception, such as authenticity, functionality, or aesthetics [3,15], lacking a comprehensive examination of how multidimensional perceived value influences purchase intention. In addition, few studies have systematically integrated the Theory of Planned Behavior (TPB) and Customer Perceived Value (CPV) to explain how consumer attitude and purchase intention are formed in the context of AI-driven fashion products.

To address this gap, this study integrates the TPB and CPV frameworks to develop a comprehensive research model that examines the key factors influencing Chinese Millennials’ willingness to accept CAGPs offered by premium fashion brands. The TPB emphasizes the predictive roles of attitude, the subjective norm, and perceived behavioral control in shaping behavioral intention, providing a theoretical foundation for understanding consumers’ willingness to adopt emerging technologies [16,17]. In contrast, CPV focuses on consumers’ evaluations of a product’s value across functional, emotional, and symbolic dimensions [18], serving as an important complement for explaining the formation of consumer attitudes. This study integrates four CPV variables—perceived aesthetic value, perceived effort of the brand in design, perceived price value, and perceived variety value—into the TPB framework. It aims to empirically test how these variables shape Chinese Millennials’ cognition, attitude, and purchase intention toward CAGPs, either indirectly through attitude formation or directly through purchase intention, thereby constructing a comprehensive explanatory mechanism. Specifically, in this study, clothing with AI-generated patterns (CAGPs) refers to garments featuring patterns created through GenAI technologies (e.g., text-to-image or image-to-image models), with human designers providing creative input alongside AI systems.

To this end, this study aims to address the identified gap by exploring the following two key questions: (1) How do attitude and subjective norm influence Chinese Millennials’ purchase intention toward CAGPs offered by premium fashion brands? (2) How do different dimensions of perceived value—such as aesthetic, emotional, experiential, and monetary—shape their purchase intention toward CAGPs, either by influencing attitude or by directly affecting purchase intention? The following section reviews the relevant literature and theoretical foundations and proposes an integrated research model. The subsequent sections present the Research Method, followed by the Results and Discussion. This research contributes to both theory and practice by extending the TPB through the lens of CPV. It explores the mechanisms linking TPB and CPV in the context of GenAI technology. In addition, it provides theoretical support for understanding Chinese Millennials’ purchase intention and optimizing value communication strategies for premium fashion brands integrating GenAI into their product offerings. Future research could further compare generational differences in perceptions of AI-driven fashion products between Millennials and Generation Z. Additionally, emerging variables such as perceived sustainability could be incorporated into the research model to expand the explanatory scope of CPV in AI-driven consumption contexts.

2. Literature Review

2.1. Emerging GenAI and Application in Fashion

Due to its powerful learning and generative capabilities, as well as its potential to enhance work efficiency, AI technology has been rapidly adopted in the fashion industry. It has accelerated the commercialization of digital fashion products on metaverse platforms [19] and digital applications such as virtual fitting rooms [20]. Moreover, new interactive formats based on GenAI, such as fashion recommendation assistants [1], have also begun to emerge. These technologies enhance consumer engagement and experiential value in the online shopping process. At the same time, consumers are increasingly paying attention to multiple value dimensions of fashion products and services supported by AI or GenAI technologies. These include design quality, aesthetic perception, experiential and hedonic value, economic value, and social value [3,15,18].

GenAI has not only expanded its applications in consumer fashion consumption and experience but has also recently emerged as a transformative tool in clothing design, particularly in the early stages of design ideation and concept development. GenAI platforms enable designers to input sketches, design themes, and material specifications to efficiently generate diverse outputs, facilitating rapid experimentation across various styles [4]. This use of GenAI also helps reduce labor and production costs while minimizing sample-related waste [15].

Despite these advantages, GenAI-generated content often lacks the aesthetic coherence and contextual awareness that characterize human design. For example, AI-generated outputs occasionally contain anatomical or stylistic distortions that conflict with consumer expectations and industry standards, such as disproportionate limbs or surreal details. These inconsistencies stem from the absence of experiential judgment and cultural sensitivity in GenAI’s decision-making process. Consumers often question the creativity of such non-human design and develop distrust toward the brand. These concerns have become key barriers to the acceptance of AI-generated fashion.

To clarify this perception gap, a conceptual comparison is presented between clothing patterns created by human designers and those generated by GenAI (Figure 1). The left side of the comparison illustrates the features of human-centered design, which include intentional creative direction, alignment with consumer preferences, and the continuous involvement of the designer throughout the process. Each stage is closely supervised to maintain aesthetic coherence and evoke emotional resonance. In contrast, the right side outlines key features of GenAI-generated patterns, including algorithm-driven pattern generation, minimal human oversight, and the occurrence of inevitable AI-generated errors. Among these, design efficiency, operational cost, brand image, consumer experience, and perceived consumer value emerge as shared factors that influence how consumers differentiate between both types of patterns. This also highlights consumers’ concerns regarding the aesthetic quality of GenAI-generated patterns, the brand’s design effort, pricing decisions, and even the speed and quantity of product launches. Overall, human designers are still needed in this stage to determine the creative direction and make detailed adjustments to CAGPs.

Given that CAGPs often retain elements of human oversight and quality control, and as brands increasingly allocate digital marketing resources toward CAGPs, consumer attention to these products continues to rise. However, the overall market acceptance of such GenAI-assisted fashion products remains contested. Prior studies have indicated that when consumers perceive insufficient human effort or a lack of emotional authenticity in AI-generated designs, their purchase intentions significantly declines [11]. In response, some brands have emphasized human involvement by foregrounding creative direction and oversight. This also reflects consumers’ growing emphasis on the origin of fashion design, as they generally believe that human designers are better able to understand and meet their nuanced needs [3].

In addition, recent studies have begun to examine the deeper implications of AI–human collaboration from an ethical perspective. As AI becomes increasingly involved in fashion design decision-making, discussions surrounding key issues such as copyright interpretability, dataset privacy, and system security have grown more frequent and urgent [21]. Consumers may also experience reduced emotional resonance with a brand’s essence and the symbolic meaning of its products when creations are perceived as non-human. This concern is particularly salient in the context of luxury brands, where emotional value is emphasized as a core attribute [22]. In this stage, some brands have adopted GenAI to support designers and enhance design uniqueness [3], positioning AI–human co-creation as an innovative marketing strategy. However, other brands—including luxury labels—continue to emphasize the role of human designers in supervising the design process and outcomes in order to build brand credibility [22]. These findings suggest that while GenAI serves as a supportive tool for fashion innovation, its impact on value expression, creator identity, and brand trust within the fashion consumption system requires critical reconsideration.

In sum, the application of GenAI in the fashion industry and clothing design is becoming increasingly widespread. However, consumers continue to weigh its benefits and drawbacks across multiple dimensions, including perceived value, emotional experience, need fulfillment, and ethical considerations. Existing studies have yet to reach a consistent conclusion. One perspective suggests that the use of GenAI may dilute consumers’ perception of the emotional value and brand essence traditionally associated with fashion brands [22]. In contrast, other research has found that customized CAGPs can enhance perceived hedonic value and foster a positive willingness to pay [9]. These inconsistencies underscore the need for further investigation into the potential factors influencing consumer attitude and purchase intention toward CAGPs.

2.2. Theoretical Background and Research Model

The Theory of Planned Behavior (TPB) [16] is one of the most widely applied frameworks in consumer behavior research. It proposes that three core factors—attitude, the subjective norm, and perceived behavioral control—primarily determine individual behavior. Specifically, attitude refers to an individual’s evaluation of potential behavioral outcomes based on personal beliefs, while the subjective norm reflects perceived social expectations from others [23].

Over the years, studies applying the TPB have expanded their focus from online shopping behavior [24] to consumer online purchasing supported by AI technologies [25], and from luxury fashion products [26] to the NFT metaverse [27] and AI-powered products [23]. Although few studies have directly applied the TPB to the field of AI-generated fashion, the aforementioned research demonstrates that the TPB framework has been widely used in related domains such as AI-enabled services, digital fashion, and AI-powered products. These studies also confirm that attitude and the subjective norm remain key predictors of behavioral intention. Moreover, the research perspectives are not limited to a global scope or Western contexts but also extend to studies focusing on East Asian regions. Therefore, these findings support the applicability of the TPB in AI-based fashion consumption contexts and its utility in predicting Chinese consumers’ behavioral intention toward CAGPs.

Although the TPB model emphasizes the predictive power of behavioral intention, it does not fully capture complex scenarios involving social interactions and emotional dynamics. To address these limitations, several extensions have been proposed, including the Affective–Cognitive TPB [28], Habitual TPB [29], and Moral–Emotive TPB [30], each integrating additional psychological or contextual dimensions. The Affective–Cognitive extension of the TPB incorporates emotional factors into the decision-making process, addressing the original model’s focus on rational cognition. For example, Razzaq et al. [31] illustrated that fashion consciousness and environmental concerns—both affective and cognitive influences—can significantly shape sustainable clothing choices. However, measuring the interaction between affect and cognition in CAGPs remains challenging. The Habitual TPB posits that certain behaviors are driven more by automatic routines than by conscious deliberation. However, its applicability to emerging technologies such as GenAI remains limited due to the novelty and low habitual familiarity of such interactions. The Moral–Emotive TPB integrates ethical considerations and emotional responses into behavioral prediction. However, moral judgments do not consistently lead to actual behavior, especially in situations where consumers experience uncertainty or are influenced by social referents [32]. Moreover, these extended TPB models do not directly account for consumers’ value perceptions and evaluations, even though such evaluations often precede and shape the formation of attitude. In fashion consumption contexts that emphasize product quality and value, assessing consumers’ multidimensional value perceptions is particularly critical—a point that has been reaffirmed by recent research on AI-generated fashion [9].

Customer Perceived Value (CPV) refers to consumers’ overall evaluation of the utility of a product or service. It is conceptualized as a multidimensional value framework that encompasses performance, price, emotional satisfaction, cognitive evaluation, and social value [18]. CPV has been shown to explain users’ emotional responses and attitudes toward emerging technological products [33], digital fashion from luxury brands [18], and AI-based products [34]. These studies confirm that various value dimensions—such as hedonic, economic, social, functional, and epistemic value—are robust predictors of consumer attitude. CPV has also been shown to effectively explain user evaluations and purchase intentions within Chinese and Southeast Asian cultural contexts. Existing studies have examined Chinese consumers’ intentions to engage in AI-supported online shopping [35], the behavioral motivations of Indonesian consumers on livestreaming platforms [36], and the mechanisms underlying Thai Millennials’ purchase intentions for luxury fashion products [37]. Therefore, by integrating CPV into the TPB framework, this study not only explains how Chinese consumers’ perceived value shapes their attitude and behavioral intention but also how attitude and the subjective norm predict purchase intention. This addresses the structural limitation of TPB in accounting for the antecedents of consumer attitude and extends the explanatory power of the model in understanding purchase intention toward CAGPs.

Specifically, this study incorporates four perceived value variables from the CPV framework, perceived aesthetic value (emotional and symbolic value), perceived price value (monetary value), perceived variety value (experiential and symbolic value), and perceived effort of the brand in design (emotional value), to explain Chinese Millennials’ attitudes and purchase intentions toward CAGPs. These dimensions have been widely validated in the contexts of emerging fashion categories, digital fashion consumption, and AI-based products [11,34,38,39]. At the same time, this study does not assume that the four perceived value variables exert identical effects. Instead, they are theoretically linked to different TPB constructs based on their underlying meanings and functional roles. Perceived aesthetic value reflects the emotional and symbolic value derived from the visual appeal of CAGPs, influencing consumer attitude through their perception of aesthetic value [40]. Perceived effort of the brand in design (PEBD) refers to consumers’ perception of the level of designer involvement and the degree of professional design expertise integrated into the development of CAGPs. As a dimension of perceived emotional value, PEBD influences consumer attitude and purchase intention through the emotional resonance and trust elicited by perceived brand effort [41]. Perceived variety value, as a dimension of experiential and symbolic value, reflects the novelty and enjoyment consumers derive from the diverse options offered by CAGPs. It fosters a positive attitude [42] and influences purchase intention by satisfying consumers’ desire for uniqueness. Perceived price value reflects monetary value, and consumers’ evaluation of the price fairness of CAGPs directly influences their purchase intention [43]. This differentiated mechanism allows each perceived value variable to enhance the TPB’s explanatory power through distinct psychological pathways. At the same time, it preserves the model’s original logic in predicting purchase intention.

The social value dimension of CPV conceptually corresponds to the TPB construct of the subjective norm, as both reflect the influence of perceived social approval on individual behavior. However, they represent different underlying psychological mechanisms. The subjective norm is a core component of the TPB. It reflects perceived social pressure—namely, an individual’s belief about whether important others think that they should perform a given behavior—and serves as a direct predictor of behavioral intention. In contrast, perceived social value focuses more on the symbolic perception of social identity that a product conveys to consumers, such as social status and group affiliation [18]. This differs from the core focus of the present study, which lies in the influence mechanism of consumers’ perceived normative social pressure. Moreover, given that this study adopts the TPB as its core theoretical framework, the use of the subjective norm helps maintain the structural integrity and theoretical coherence of the model. Prior research has also demonstrated that normative influence—specifically the subjective norm—plays a crucial role in shaping behavioral intention among Chinese Millennials in the context of premium fashion consumption [44]. Therefore, this study adopts the subjective norm as the social influence pathway to ensure theoretical consistency within the TPB framework. This choice aligns more clearly with the model’s structural logic and the study’s research objectives. Furthermore, in this research framework, attitude specifically refers to consumers’ integrated emotional and cognitive evaluation of CAGPs offered by premium fashion brands. It serves as a key psychological mechanism and mediating variable that links perceived value dimensions to purchase intention.

Previous studies have successfully applied the integrated TPB–CPV framework to investigate user behavior or purchase intention in various domains, including emerging technological products [45], sustainable brand offerings [46], digital technologies [17], and online retail [47]. These studies demonstrate the value of this integrated model in explaining complex purchase intentions that involve both emotional and rational trade-offs. Therefore, this study primarily aims to provide a comprehensive understanding of how internal value evaluation (attitude) and external social influence (subjective norm) jointly shape consumer responses to CAGPs. In doing so, it also strengthens the theoretical rationale for incorporating perceived value dimensions into attitude formation within the TPB framework.

2.3. Millennial Consumers, Premium Fashion, and Hypothesis Development

Understanding consumer motivations is critical for brand managers, particularly when engaging tech-savvy demographic groups. Millennials, now firmly established in the workforce, are reshaping the global apparel market through their distinct behaviors, preferences, and values [48]. Compared to Generation X, they are more receptive to emerging fashion trends. This makes them a strategically important consumer segment for premium fashion brands introducing CAGPs in dynamic online retail settings. Accordingly, aligning product strategies with Millennial consumption patterns requires a deeper understanding of the factors that influence their purchase intentions.

Prior studies have identified various determinants of Millennials’ online shopping behavior, including product awareness [46], fashion involvement [49], customer satisfaction [50], and risk perception. In the context of digital technology adoption, factors such as perceived usefulness and novelty [51] have been shown to significantly influence their willingness to adopt AI technologies. When purchasing premium goods online, Millennials exhibit increased sensitivity to price and product availability [52], and they place particular emphasis on peer reviews and electronic word of mouth [7].

The above literature suggests that Millennials’ behavioral intentions are shaped by product attributes (objective) and consumer perceptions (subjective). As premium brands target the middle class, including Millennials with purchasing power and willingness to spend, offering high-quality, aesthetically distinct products [52] becomes especially critical in meeting their expectations for quality and brand identity. Based on the above evidence, it can be observed that the consumer psychology of Millennials reflects their perception of value. Specifically, they seek product finish and design quality [53], appreciate design aesthetics [54], are sensitive to pricing [55], care about brand image [56], pay attention to social evaluation [52], and pursue product variety as a means of expressing a unique identity [40]. These factors collectively highlight the perceived value frameworks underpinning Millennials’ online purchase intentions—particularly in contexts involving CAGPs. Their decision-making reflects a complex interplay between cognitive evaluations, emotional responses, and behavioral intentions.

These value orientations have also begun to attract attention in studies on the acceptance of AI-based fashion technologies and services in China and Southeast Asia. For example, Shin and Yang [12] found that perceived enjoyment significantly enhanced Chinese consumers’ attitudes toward using AI fashion services. Gao and Liang [14] empirically demonstrated that both utilitarian and hedonic value play a positive role in shaping young Chinese consumers’ impulsive purchase intentions. These values are derived from AI-based virtual try-on technologies, which enhance both functional and experiential aspects of the shopping process. In related research from Southeast Asia, Yeo et al. [57] also identified the emotional value and perceived quality of AI-based digital technology experiences as significant factors influencing fashion purchase decisions among Malaysian Instagram users. These studies suggest that the application of AI in the fashion industry is not only a global trend but also a research topic grounded in real market potential across China and Southeast Asia. However, existing research on purchase intentions toward GenAI-based fashion products remains largely centered on Western perspectives [15,22]. Consumers’ value perceptions and purchase intentions are shaped by cultural context. This is particularly relevant in China, an emerging market where AI technologies are being rapidly adopted. However, how different dimensions of perceived value influence Chinese Millennials’ attitudes and purchase intentions toward CAGPs remains underexplored. The following section, based on a review of the relevant literature, proposes a hypothesized research model and a set of hypotheses (see Figure 2) to explore these mechanisms in greater depth.

Although the predictive variables adopted in this study—such as perceived price value, the subjective norm, and attitude—are derived from general consumer behavior models, they remain theoretically applicable in the context of AI-generated fashion. Prior research [58,59] has shown that general variables such as the subjective norm, attitude, and perceived monetary value can be effectively applied to AI-generated contexts when their measurement items are adapted to specific use scenarios. This approach allows for an effective understanding of how users cognitively and behaviorally respond to AI-generated creative content.

2.3.1. Perceived Aesthetic Value

The perceived aesthetic value (PAV) of clothing refers to consumers’ subjective evaluation of visual characteristics, such as style, color, and patterns [38]. These attributes represent consumers’ perception of aesthetic value and are highly valued by them [40]. As a result, aesthetic value has become a primary consideration for designers, making it a key factor influencing consumer choices and purchase decisions. PAV reflects not only the visual appeal of a product but also the emotional and symbolic value derived from aesthetically pleasing designs.

The existing literature highlights the critical role of PAV in shaping consumer attitudes toward clothing purchases. For example, Hwang et al. [38] demonstrated a significant influence of aesthetic value on American consumers’ attitudes toward purchasing solar-powered apparel. Similarly, Shi et al. [54] found high aesthetic quality to evoke positive emotional responses, thereby enhancing consumer acceptance.

In the context of AI-generated clothing patterns, consumer attitudes and purchase intentions may depend on whether these designs’ aesthetic features align with their visual expectations. Accordingly, we propose the following hypothesis:

H1. 

The PAV of CAGPs positively affects Chinese Millennials’ attitudes toward such products offered by premium fashion brands.

2.3.2. Perceived Effort of the Brand in Design

Effort is often used as a heuristic for quality assessment: the more effort invested in development and production, the better the product is perceived to be [60]. Perceived effort of the brand in design (PEBD) refers to consumers’ perception of the effort a brand invests in product design. This primarily includes the degree of designers’ involvement, as well as their professional knowledge and experience throughout the creative process. In AI-assisted design, cost-effectiveness must be balanced with consumer expectations. Excessive reliance on automation may undermine brand authenticity and diminish emotional trust, deterring potential CAGP buyers. When consumers perceive that a premium fashion brand adopts GenAI technology while maintaining strong designer involvement to ensure design quality, they are more likely to recognize the brand’s effort and experience positive emotions [61]. This may lead to emotional identification, such as the belief that “the brand is attentive to design”. Thus, PEBD is considered part of consumers’ emotional value perception, a key component of CPV [62].

Fashion design typically requires intensive information processing and creative exploration, demanding considerable time and cognitive effort. This perceived designer involvement fosters consumer trust and brand loyalty, often reinforced through repeated purchase behavior [41]. However, prior studies indicate that the introduction of AI-led designs may significantly reduce consumers’ purchase intentions [11]. AI-generated patterns are often perceived as the result of limited creative effort, involving minimal human intellectual input. Consequently, consumers may perceive that premium fashion brands have reduced their investment in design. Based on the above, we propose the following hypotheses:

H2. 

PEBD positively influences Chinese Millennials’ attitudes toward CAGPs from premium fashion brands.

H3. 

PEBD positively influences Chinese Millennials’ online purchase intentions for CAGPs from premium fashion brands.

2.3.3. Perceived Variety Value

Perceived variety value (PVV) refers to consumers’ perceived value derived from the number and diversity of options available within a product or service category [42]. In fashion consumption, variety enables consumers to express individuality and distinguish themselves from others [40], while also enhancing emotional and cognitive utility through novelty, enjoyment, and the freedom of self-expression [42]. As such, PVV captures both the experiential and symbolic value associated with having a broader range of stylistic choices.

Research has shown that greater perceived variety is associated with positive emotions, favorable evaluations, and stronger behavioral intentions. In online retail contexts, variety significantly enhances consumers’ perceptions of product quality, thereby supporting more positive purchasing decisions [63]. For instance, Chang [39] demonstrated that PVV enhances both utilitarian and hedonic satisfaction in digital shopping environments. Increasingly, brands and designers leverage GenAI to generate diverse clothing patterns across various styles and themes, enabling them to better satisfy consumers’ desires for uniqueness and trend alignment [64]. This capability is especially salient for Millennial consumers, who actively pursue novelty and variety in fashion products [65]. Given this context, the following hypothesis is proposed:

H4. 

PVV positively influences Chinese Millennials’ attitudes toward CAGPs from premium fashion brands.

H5. 

PVV positively influences Chinese Millennials’ online purchase intentions for CAGPs from premium fashion brands.

2.3.4. Perceived Price Value

Perceived price value (PPV) refers to consumers’ evaluation of whether a product’s price is reasonable relative to the value they expect to obtain [66]. As a component of the monetary value dimension within Customer Perceived Value (CPV), it plays a direct role in shaping purchasing behavior [67].

In online fashion retail, price serves as a key heuristic by which consumers infer product quality [63]. Prior research has shown that perceived price significantly influences purchasing decisions [43], particularly among Millennials, who actively compare prices and seek affordable luxury options [55]. While they are willing to invest in premium products, price sensitivity remains a constraint, with thresholds, such as the USD 250 benchmark, potentially deterring purchase intention [55].

GenAI technology enables the rapid and cost-efficient generation of fashion patterns, raising questions about whether these cost savings are reflected in product pricing. As consumers become increasingly aware of such technological advantages, they may expect CAGPs to be reasonably and transparently priced, offering value for money. Ultimately, PPV captures how consumers weigh perceived benefits against the price paid, and this price–value trade-off may become a decisive factor in Millennials’ purchase decisions. Based on this reasoning, the following hypothesis is proposed:

H6. 

PPV positively influences Chinese Millennials’ online purchase intentions for CAGPs from premium fashion brands.

2.3.5. Subjective Norm

The subjective norm (SN) refers to an individual’s perception of social pressure from important others, such as family and peers, regarding whether they should perform a particular behavior [68]. Studies suggest that Millennials often rely heavily on peer approval in their fashion choices [7], while a lack of social support may reduce their willingness to adopt emerging fashion trends [69]. Among young consumers, Chinese Millennials, sensitivity to external social expectations tends to be more pronounced [70].

Prior research has shown that the SN positively influences purchase intention across various contexts [32]. For instance, Kang et al. [44] found that Chinese Millennials’ luxury brand consumption was significantly shaped by the opinions of surrounding consumers. According to the TPB, more favorable social norms are associated with stronger behavioral intentions [16].

As a novel form of fashion expression, AI-generated clothing patterns embody both technological innovation and evolving aesthetic preferences. Since clothing serves as a means of social identity construction—particularly among younger consumers—social evaluations play a critical role in shaping purchase behavior [69]. Accordingly, we propose the following hypothesis:

H7. 

The SN positively influences Chinese Millennials’ online purchase intentions for CAGPs from premium fashion brands.

2.3.6. Attitude—Purchase Intention Relationship

In the TPB, attitude is the core predictor of behavioral intention [46]. In this study, attitude refers to consumers’ overall evaluation of the aesthetic, symbolic, and hedonic values of CAGPs launched by premium fashion brands. Attitude captures consumers’ emotional responses [71], which are formed after assessing various value dimensions of CAGPs, such as aesthetic appeal, price reasonableness, and product variety.

The positive relationship between consumer attitudes and purchase intentions has been widely supported in fashion-related studies, including research on sustainable apparel [68] and leather fashion products [49]. Bian and Forsythe [72] further demonstrated that consumers’ social and functional attitudes toward luxury brands positively influence their purchase intentions by eliciting emotional responses.

As a relatively new concept, CAGPs remain unfamiliar to many consumers. Nevertheless, Millennials generally hold favorable attitudes toward innovations in fashion, including products driven by GenAI. These positive attitudes are likely to be reflected in their purchase intentions. Therefore, the following hypothesis is proposed:

H8. 

Chinese Millennials’ attitudes toward CAGPs from premium fashion brands positively influence purchase intentions.

3. Research Method

3.1. Questionnaire and Instruments

This study explores the factors influencing Chinese Millennials’ attitudes and purchase intentions toward online consumption with CAGPs from premium fashion brands. The questionnaire in this study was adapted from previously validated multi-item English scales and modified to fit the research context. The translation process followed a bilingual back-translation procedure to ensure linguistic accuracy and cultural appropriateness. The initial translation and back-translation were conducted separately by two bilingual experts with academic backgrounds in fashion management and consumer behavior. A cross-checking process was then implemented. Ambiguous items were revised to ensure clarity, and efforts were made to eliminate any redundancy or overlap among the measurement items.

This study employed a seven-point Likert scale (1 = “strongly disagree”; 7 = “strongly agree”) to capture respondents’ perceptions from multiple dimensions and to accurately reflect subtle differences in their responses to each item [73]. A bipolar semantic differential scale was used to measure attitude. This method helped capture the direction and intensity of respondents’ emotional evaluations of the multidimensional value reflected in CAGPs. This measurement approach is widely used in psychometric scales assessing consumers’ overall evaluations of products [15,18], as it prompts participants to consider specific psychological dimensions (e.g., undesirable–desirable), leading to more authentic, accurate, and nuanced responses.

To ensure a consistent understanding of the CAGP concept throughout the survey, participants were shown a set of visual stimuli before answering the questionnaire. These stimuli consisted of examples of patterns designed with GenAI tools by premium fashion brands. This approach helped reduce subjective cognitive bias caused by the lack of a unified standard among participants, especially when responding to items related to perceived aesthetic value.

In addition, to ensure that the measurement items accurately reflected participants’ evaluations of CAGPs, this study also referred to practical approaches in the relevant literature. Specifically, the questionnaire items were contextually tailored and the wording was adjusted to fit the specific research setting [15,58,59]. This approach effectively integrated general variables (such as perceived aesthetic value, attitude, and subjective norm) into the evaluation context of CAGPs, for example, “AI-generated clothing patterns from premium fashion brands match my personal aesthetic preferences”. By instructing respondents to focus on “CAGPs launched by premium fashion brands” as the object of evaluation, this study enhanced the content validity of the general variables. Details of the measurement items in

Table 1. Summary of constructs and measurement items.

Variables	Items	Code
Perceived Aesthetic Value (PAV)	AI-generated patterns on premium brand clothing are visually appealing.	PAV1
	AI-generated clothing patterns from premium brands match my personal aesthetic preferences.	PAV2
	AI-generated clothing patterns from premium brands can be trendy and well-designed.	PAV3
Perceived Effort of Brand in Design (PEBD)	The brand puts thoughtful effort into pattern design, even using AI-generated designs.	PEBD1
	AI-generated patterns from premium brands reflect the same level of design effort as human-created ones.	PEBD2
	Using AI-generated patterns may give the impression that the brand invested less design effort.	PEBD3
Subjective Norm (SN)	People who are important to me think I should purchase premium-brand CAGPs.	SN1
	People whose opinions I value would approve of me purchasing premium-brand CAGPs.	SN2
	Social norms influence my intention to purchase premium-brand CAGPs.	SN3
Perceived Price Value (PPV)	AI-generated clothing patterns should be priced fairly, considering the use of automated technology.	PPV1
	I find the pricing of premium-brand CAGPs reasonable, even if the patterns are AI-generated.	PPV2
	AI-generated clothing patterns offer good value for money.	PPV3
Perceived Variety Value (PVV)	I feel there is a wide selection of CAGPs available from premium brands.	PVV1
	The CAGPs provide a diverse range of style options.	PVV2
	CAGPs are suitable for a variety of occasions and dressing needs.	PVV3
Attitude Toward Online Purchase of CAGPs from Premium Brands (AT)	If a premium brand announces that it has used GenAI technology to assist in designing clothing patterns, my evaluation of the product is:
	Unappealing–Appealing.	AT1
	Bad–Good.	AT2
	Unfavorable–Favorable.	AT3
	Undesirable–Desirable.	AT4
	Worthless–Valuable.	AT5
Purchase Intention (PI)	I intend to purchase CAGPs from premium brands in the near future.	PI1
	I am likely to choose clothing with AI-generated patterns when shopping from premium brands.	PI2
	If I find CAGPs from premium brands online, I would consider purchasing them.	PI3
	I am willing to pay for CAGPs offered by premium brands.	PI4
	I would actively seek opportunities to purchase CAGPs from premium brands.	PI5

.

Perceived aesthetic value (PAV) was measured using items adapted from Hwang et al. [38], who examined consumers’ aesthetic needs in the context of smart clothing. The items were reformulated to reflect the visual characteristics of AI-generated patterns in fashion. Perceived effort of the brand in design (PEBD) was measured using three items adapted from Grewal and Stephen [74], originally developed to capture perceived brand effort in digital retail settings. These items were modified to emphasize design input in an AI-assisted fashion context. Perceived variety value (PVV) was measured using three items adapted from Chang [39], focusing on consumers’ perceptions of product diversity and assortment in online fashion. Perceived price value (PPV) was measured using three items adapted from Lichtenstein et al. [66], reflecting consumer evaluations of price fairness and value for money in CAGPs. The subjective norm (SN) were measured with three items adapted from Rausch and Kopplin [68], capturing perceived social influence from peers and important others regarding CAGP purchase decisions. Attitude toward the online purchase of CAGPs from premium fashion brands was measured using five items adapted from Rausch and Kopplin [68], reformulated to assess emotional and cognitive evaluations specific to AI-generated fashion. Purchase intention (PI) was measured using five items adapted from Sohn and Kwon [23], focusing on consumers’ willingness to purchase CAGPs from premium fashion brands in the near future.

A pre-test was conducted with 25 fashion design students to assess the questionnaire’s clarity, contextual relevance, and content validity. Based on the responses, redundant or ambiguous items (e.g., “When I purchase clothing, I will primarily consider CAGPs”) were removed or rephrased to improve readability and alignment with this study’s context. The pre-test showed strong internal consistency, with a Cronbach’s alpha of 0.80 for the overall scale and values ranging from 0.79 to 0.93 across individual constructs. Given these satisfactory results, the finalized questionnaire was used for large-scale data collection and empirical analysis.

3.2. Sampling and Data Collection

Data were collected via Credamo, a professional online research and modeling platform. The questionnaire was distributed online from 12 October to 18 November 2024, with an average completion time of 4 to 6 min. Respondents received a small reward in RMB upon completion. The survey lasted for 38 days, allowing sufficient time for data collection. A purposive sampling strategy was adopted to align with the research objectives. The target group consisted of Chinese Millennials aged 29 to 44 residing in the economically advanced Jiangsu–Zhejiang–Shanghai region. This study selected this region as the sampling area due to its vibrant fashion industry, well-developed e-commerce ecosystem, and strategic advantages as a hub for AI technology clusters [75]. Millennials in this region are more frequently exposed to digital and AI-driven fashion compared to those in other areas, which subtly influences their consumption values and preferences. According to the 2024 China E-commerce Industry Report [76], Millennials in the Jiangsu–Zhejiang–Shanghai region lead compared to their counterparts in other regions of China in terms of online shopping amount. They also surpass their peers in shopping frequency. Millennials in the Jiangsu–Zhejiang–Shanghai region demonstrate both a willingness to consume premium brands [77] and an open attitude toward digital innovation. Therefore, they are likely to provide particularly relevant insights when evaluating CAGPs launched by premium fashion brands.

A total of 608 responses participated, resulting in a response rate of 93.25%. After excluding incomplete responses and those with a completion time of less than 180 s, 471 valid responses were retained for analysis, yielding a valid response rate of 77.5%. Considering the complexity of the research model and the number of measurement items for each predictor variable, the final valid sample size of 471 far exceeded the minimum sample size recommended by Hair et al. [78] for structural equation modeling (SEM). This recommendation, known as the “10 times rule,” suggests a sample size that is at least ten times the number of paths pointing to the most complex dependent construct. Therefore, the number of valid responses collected in this study provides sufficient statistical power to support both the measurement model and hypothesis testing.

Among the respondents, 51.8% were male and 48.2% were female. In terms of educational background, 28.66% had completed high school or vocational education, 50.96% held a bachelor’s degree, and 20.36% had a master’s degree or higher. As for monthly disposable income, 31.00% reported earnings between RMB 4000 and 6000, 17.41% between RMB 6000 and 10,000, and 8.70% between RMB 10,000 and 15,000.

3.3. Data Analysis

Data analysis involved a series of steps, including descriptive analysis, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and structural equation modeling (SEM), conducted using SPSS version 26.0 and AMOS version 27.0. Descriptive statistics were used to profile the respondents and evaluate data quality. The skewness and kurtosis values of all scales were examined to assess the normality of data distribution and its suitability for parametric analysis. Exploratory factor analysis was conducted using principal axis extraction and Varimax rotation on all measurement scales. Subsequently, a CFA using maximum likelihood estimation was performed on the 25 scale indicators. Given the successful levels of reliability and validity of the measurements, the fit of the structural model was estimated to test the proposed research model and hypotheses.

4. Results

4.1. Descriptive Statistics

Table 2. Measurement items and psychometric properties of constructs.

Constructs	Items	M	SD	Cronbach’s Alpha	AVE	CR	EFA Loading	CFA Loading
Perceived Aesthetic Value	PAV1	5.32	1.15	0.78	0.52	0.76	0.78	0.70
	PAV2	5.26	1.18				0.78	0.74
	PAV3	5.13	1.23				0.71	0.72
Perceived Effort of Brand in Design	PEBD1	4.03	1.46	0.86	0.69	0.87	0.85	0.86
	PEBD2	3.86	1.43				0.84	0.88
	PEBD3	4.39	1.38				0.68	0.74
Subjective Norm	SN1	5.01	1.20	0.82	0.63	0.83	0.82	0.78
	SN2	5.02	1.21				0.87	0.92
	SN3	4.64	1.34				0.80	0.66
Perceived Price Value	PPV1	4.39	0.99	0.76	0.51	0.76	0.76	0.69
	PPV2	4.38	0.98				0.82	0.79
	PPV3	4.23	1.04				0.73	0.66
Perceived Variety Value	PVV1	4.51	1.25	0.85	0.66	0.85	0.81	0.74
	PVV2	4.59	1.29				0.87	0.87
	PVV3	4.65	1.33				0.81	0.82
Attitude Toward Online Purchase of CAGPs from Premium Fashion Brands	AT1	4.58	1.27	0.88	0.59	0.88	0.69	0.72
	AT2	4.45	1.32				0.73	0.76
	AT3	4.14	1.36				0.87	0.81
	AT4	4.41	1.41				0.78	0.79
	AT5	4.14	1.40				0.83	0.75
Purchase Intention	PI1	4.80	1.33	0.88	0.63	0.87	0.69	0.71
	PI2	4.52	1.51				0.82	0.83
	PI3	4.62	1.36				0.73	0.77
	PI4	4.65	1.23				0.85	0.85

Note: M = mean; SD = standard deviation; CR = composite reliability; AVE = average variance; EFA = exploratory factor analysis; CFA = confirmatory factor analysis.

presents the descriptive statistics for all measured constructs. The mean scores of the 24 items across the seven constructs ranged from 3.86 to 5.32, indicating that respondents generally expressed neutral to moderately positive evaluations. Standard deviations ranged from 0.98 to 1.51, suggesting acceptable variability in the data. Additionally, all skewness and kurtosis values fell within the commonly accepted threshold of ±2 [79], supporting the assumption of univariate normality for subsequent parametric analyses.

4.2. Measurement Model Evaluation and Common Method Bias Assessment

This study tested for common method bias (CMB) because all measurement items were completed by the same respondents. Harman’s single-factor test was used for this purpose. The results showed that the single extracted factor explained 33.60% of the total variance, which did not exceed the 50% threshold set by Harman’s test [80]. Therefore, the risk of common method bias in this study is low and is unlikely to significantly affect the validity of the results.

In the EFA results, a seven-factor model was identified. After examining communality values, factor loadings, and cross-loadings, one item—“If I come across an online store of a premium fashion brand selling AI-generated pattern clothing, I would plan to browse and purchase such clothing”—was removed due to its low communality (0.35). The final analysis showed that 24 items loaded onto seven factors, explaining 73.46% of the total variance. Communality values ranged from 0.63 to 0.84, and the Kaiser–Meyer–Olkin (KMO) measure was 0.89. Table 2 presents the EFA loadings, which ranged from 0.68 to 0.87.

SEM allows for the simultaneous examination of multiple variables, accounts for measurement error, and integrates both factor and path analysis, making it more robust than traditional regression. CFA in AMOS version 27.0 was conducted to assess validity, reliability, and model fit, and to remove items with low standardized factor loadings. The measurement model (Figure 3) was tested, and all factor loadings were statistically significant (p < 0.001). Model fit indices were acceptable, indicating a good overall fit: CMIN/DF = 3.10; CFI = 0.92; GFI = 0.88; AGFI = 0.84; TLI = 0.90; NFI = 0.88; and RMSEA = 0.07. Given the sample size and model complexity, a CMIN/DF value between 3 and 5 was considered acceptable. Cronbach’s alpha coefficients ranged from 0.76 to 0.88, indicating satisfactory internal consistency (see Table 2), and confirming that the variables were reliable and appropriate for further analysis.

Additionally, the measurement model was assessed for convergent validity, discriminant validity, and internal reliability. The CFA loadings ranged from 0.66 to 0.92, indicating acceptable item reliability. The Average Variance Extracted (AVE) for each construct ranged from 0.51 to 0.69 (see Table 2), exceeding the 0.50 threshold. This suggests that each construct explained more than half of the variance in its indicators, thus supporting convergent validity [18]. The square root of the AVE for each construct was greater than its highest correlation with any other construct (see

Table 3. Discriminant validity and correlation matrix.

	Perceived Aesthetic Value	Perceived Effort of Brand in Design	Subjective Norm	Perceived Price Value	Perceived Variety Value	Attitude Toward Online Purchase of CAGPs from Premium Fashion Brands	Purchase Intention
Perceived Aesthetic Value	0.52 (0.72)
Perceived Effort of Brand in Design	0.32 **	0.69 (0.83)
Subjective Norm	0.29 **	0.37 **	0.63 (0.79)
Perceived Price Value	0.34 **	0.41 **	0.18 **	0.51 (0.72)
Perceived Variety Value	0.36 **	0.37 **	0.25 **	0.37 **	0.66 (0.81)
Attitude Toward Online Purchase of CAGPs from Premium Fashion Brands	0.47 **	0.36 **	0.15 **	0.34 **	0.30 **	0.59 (0.77)
Purchase Intention	0.45 **	0.57 **	0.35 **	0.45 **	0.36 **	0.43 **	0.63 (0.79)

Note: Bold values in the diagonal represent the AVE of each construct, and the square roots of AVE are shown in parentheses below in italics. Off-diagonal values are Pearson correlation coefficients. Double asterisks (**) indicate that the correlation is significant at the 0.01 level (2-tailed). Discriminant validity is confirmed when the square root of AVE exceeds inter-construct correlations (Fornell–Larcker criterion).

). This indicates that each construct shares more variance with its own indicators than with other constructs, confirming discriminant validity.

4.3. Structure Model and Hypotheses Testing

The structural equation model (Figure 3) was used to test the proposed hypotheses. The model proposed that PAV, PEBD, and PVV would influence consumers’ attitudes toward CAGPs. In addition, PEBD, SN, PPV, PVV, and attitude were hypothesized to have direct effects on purchase intention.

The model testing results are presented in

Table 4. Standardized path coefficients and significance levels from the structural model.

Dependent Variable	Hypothesis	Path	Standardized Path Coefficient	S.E.	C.R.	p	Hypothesis Supported
AT	H1	PAV→AT	0.46	0.07	7.31	***	Yes
	H2	PEBD→AT	0.21	0.04	4.18	***	Yes
	H4	PVV→AT	0.06	0.04	1.08	0.28	No
PI	H8	AT→PI	0.21	0.05	4.36	***	Yes
	H3	PEBD→PI	0.40	0.05	7.45	***	Yes
	H7	SN→PI	0.17	0.05	3.43	***	Yes
	H6	PPV→PI	0.26	0.07	5.15	***	Yes
	H5	PVV→PI	0.06	0.04	1.31	0.19	No

Note: *** p < 0.001.

. PAV had a significant positive effect on attitude (β = 0.46; p < 0.001), supporting H1. PEBD also had positive effects on both attitude (β = 0.21; p < 0.001) and purchase intention (β = 0.40; p < 0.001), confirming H2 and H3. SN positively influenced purchase intention (β = 0.17; p < 0.001), supporting H7. Similarly, PPV had a significant positive effect on purchase intention (β = 0.26; p < 0.001), providing support for H6. Attitude also significantly influenced purchase intention (β = 0.21; p < 0.001), confirming H8.

In contrast, PVV had no significant effect on either attitude or purchase intention (p > 0.05), and thus, H4 and H5 were not supported. Among all significant predictors, PAV exerted the strongest effect on attitude. PEBD, meanwhile, showed the most substantial direct influence on purchase intention. A summary of all tested paths is presented in Table 4.

4.4. Mediation Analyses

The bias-corrected bootstrap method was used to examine the mediating effect of attitude, following the procedure recommended by MacKinnon et al. [81]. A total of 5000 bootstrap samples were generated to estimate indirect effects and construct 95% confidence intervals, without assuming normality in the sampling distribution.

The results (

Table 5. Mediation analysis results: indirect path coefficients.

Path		Effect	Bias-Corrected 95% CI		Result
			Lower Bound	Upper Bound
PEBD→AT→PI	Total Effects	0.45	0.35	0.55	Partial mediation
	Direct Effects	0.40	0.30	0.51
	Indirect Effects	0.04	0.02	0.08
PVV→AT→PI	Total Effects	0.07	−0.02	0.16	Unclear whether mediation effects exist
	Direct Effects	0.06	−0.04	0.15
	Indirect Effects	0.01	−0.01	0.04

) show that the indirect effect of PEBD on purchase intention through attitude was statistically significant, with a confidence interval excluding zero. This indicates that attitude partially mediates the relationship between PEBD and purchase intention. In contrast, for the path from PVV to purchase intention via attitude, the confidence interval included zero, suggesting that PVV did not exhibit a significant indirect effect on purchase intention through attitude.

5. Discussion and Implications

5.1. Discussion

This study found that perceived aesthetic value significantly enhances consumer attitude toward CAGPs. However, unlike Zhang and Liu [15], who reported that aesthetic value did not significantly influence consumer acceptance of AI-designed hemp fashion products, our findings underscore its relevance. Their study emphasized expressiveness—comprising communicative value, symbolic value, and emotional value—as the primary driver. Our findings partially align with this view by confirming the role of emotional value: perceived brand effort in design was shown to positively affect both attitude and purchase intention, supporting the findings of Li et al. [11]. These results suggest that future research should further explore how aesthetic value and emotional value interact in shaping consumer attitude. These results were observed in the context of premium fashion consumption. They support the theoretical assumptions of the CPV framework, which states that perceived aesthetic value and perceived emotional value influence consumers’ attitude evaluations [54,62].

Furthermore, we confirmed the positive impact of the subjective norm on purchase intention. This finding is consistent with the TPB framework, which identifies the subjective norm as a key predictor of behavioral intention driven by social influence [16]. This study extends its applicability to the domain of GenAI fashion products, which previous studies have not explicitly addressed. The testing results provide further insight into how social influence shapes consumer behavioral intention in this emerging context. While Tsuchiya et al. [67] found that monetary value positively affects purchase intention for “fast fashion” products, the technological distinctions between “fast fashion” and GenAI-supported fashion products suggest that their findings may not be entirely generalizable. Our study underscores the significant role of perceived price value (monetary value) in shaping purchase intention for GenAI clothing. This finding supports the CPV framework’s presumption that the monetary value component significantly influences behavioral intention, indicating that Millennials show price sensitivity when evaluating premium fashion [55], further reinforcing the relevance of monetary value in this domain.

One noteworthy finding is that perceived variety value did not significantly influence either attitude or purchase intention. This finding contrasts with previous studies on fashion e-commerce consumption [63], suggesting that perceived variety value may operate differently in the context of AI-driven fashion. Future research should examine whether this divergence is context-specific or reflects a broader shift in consumer preferences toward AI-generated fashion. The following section discusses these findings in greater detail.

5.1.1. Design Quality Factors: Drivers of Consumer Need Satisfaction

In the early stages of market promotion, the adoption of AI products is not always driven by product quality [82]; instead, visual design plays a more decisive role in attracting consumers. This study found that perceived aesthetic value significantly influences Chinese Millennials’ attitudes toward CAGPs from premium fashion brands. This result is consistent with prior studies indicating that products with strong aesthetic appeal tend to elicit positive emotional responses in consumers [54].

Meeting emotional needs within the cultural context of young Chinese consumers is essential [14]. Research shows that perceived brand effort in design, as an evaluative emotional factor, significantly influences Chinese Millennial attitude and purchase intention, supporting previous findings [11]. Premium fashion brands often adopt GenAI technology to reduce time and labor costs. However, the use of GenAI may reduce the perceived brand effort in design among Chinese Millennials. Among all variables, PEBD has the strongest effect on purchase intention, highlighting the importance consumers place on design creativity and quality.

In addition, attitude plays a partial mediating role in the relationship between perceived brand’s effort in design and purchase intention. This suggests that, for Chinese Millennials, perceived design effort not only directly stimulates their intention to purchase CAGPs but also evokes positive emotional satisfaction. Such positive feelings can lead to more favorable evaluations of the brand’s design commitment and professional capabilities, thereby positively influencing their purchase decisions. The direct and indirect effects were both significant. This reinforces the importance of perceived effort of the brand in design, as well as the creative and emotional resonance it conveys, in shaping consumers’ purchase intention toward CAGPs.

5.1.2. The True Influence of the Subjective Norm on Behavioral Intention

This study shows that subjective norms have a significant positive influence on purchase intention. Clothing, as an essential part of daily life, is traditionally regarded as a social product, and its consumption is often shaped by perceived social influence from others [68]. Our finding aligns with previous research [44], suggesting that when Millennials observe their peers or fashion leaders wearing a product, they are more likely to purchase it.

5.1.3. The Significant Impact of Perceived Price Value

Price remains a critical factor influencing consumer decision-making [66], and this also applies to how consumers evaluate CAGPs. The present study confirms that perceived price value has a significant positive impact on purchase intention. This result is consistent with previous research [67] and may be explained by the price sensitivity commonly observed among Millennials, especially when evaluating premium brands. Millennials often prioritize price when making purchase decisions [55]. When the price of CAGPs is perceived as favorable, it suggests that the product meets their value-for-money expectations, thereby enhancing their purchase intention.

5.1.4. Logical Confirmation: A Positive Correlation Between Attitude and Purchase Intention

Attitude reflects individuals’ cognitive and emotional evaluations of objects [12]. This study also confirms a significant positive relationship between Chinese Millennials’ general attitude toward CAGPs from premium fashion brands and their purchase intention. This favorable attitude may result from various social and personal behaviors shaped by the GenAI-driven fashion environment. These behaviors include the growing adoption of GenAI technology in fashion design, endorsement by major brands, experimentation by fashion pioneers, and the spontaneous spread of trends through social platforms. This finding reinforces earlier research indicating that attitude positively influences purchase intention [34,72].

5.1.5. Rethinking Perceived Variety Value in AI Fashion Contexts

This study also found that perceived variety value had no direct effect on either attitude or purchase intention. This result contrasts with previous findings [63] and challenges the common assumption that offering diverse product choices enhances consumers’ perceived experiential value. This assumption is particularly prevalent in fashion consumption contexts, where variety is often sought to increase the likelihood of matching personal preferences. The indirect path from perceived variety value to purchase intention through attitude was also not significant. Together, these findings suggest that perceived variety value does not effectively predict the emotional responses or purchase intention of Chinese Millennials in the context of CAGPs. One possible explanation for this result is that, although using AI to generate clothing patterns can improve design efficiency and provide opportunities to expand product lines and style variations, Chinese Millennials may perceive AI as merely “stitching together creativity” or relying on algorithmic combinations [3]. As a result, they may view CAGPs as lacking authenticity and question the originality and novelty of the designs. Moreover, they might interpret the brand’s ability to release a larger number of styles in a short period as a sign of reduced designer effort and oversight.

Moreover, although Chinese Millennials expect to see novel and diverse offerings from premium fashion brands, diversity alone is not enough to appeal to this group. High-quality design and product uniqueness remain essential prerequisites [52]. If premium fashion brands focus solely on showcasing product variety, it may conflict with their brand image in the eyes of consumers—one that is characterized by carefully curated themes, creative design, high quality, and a sense of scarcity [58]. These attributes are particularly important in shaping the premium fashion brand experience.

5.2. Implications

5.2.1. Theoretical Implications

This study makes three key contributions to the literature on consumer behavior, technology acceptance, and fashion marketing. First, this study proposes an integrated framework combining the Theory of Planned Behavior and Customer Perceived Value to investigate the emerging field of CAGPs. This study does not adopt a parallel integration of the TPB and CPV. Instead, it incorporates diverse perceived value variables related to Chinese Millennials’ views on the design quality, brand essence, and market value of CAGPs. These variables help further explain the antecedents of attitude and purchase intention, while also linking them to the psychosocial mechanism of subjective norm. In doing so, this study refines the explanatory path from cognition to attitude and ultimately to purchase intention regarding CAGPs introduced by premium fashion brands. This study demonstrates the joint explanatory power of the TPB and CPV within a unified framework. It also offers a general theoretical foundation for future research on consumer decision-making in the context of GenAI fashion products.

Secondly, this study introduces and validates two underexplored variables—perceived brand effort in design and perceived price value—as antecedents of Chinese Millennials’ purchase intention. These variables complement the value evaluation perspective within the attitude formation process. Specifically, Chinese Millennials’ evaluations of CAGPs are not only shaped by traditional TPB variables but also influenced by premium fashion brands’ commitment to design creativity and quality in the GenAI-assisted design process, as well as the perceived reasonableness of pricing. This contributes to extending the explanatory scope of TPB in the context of GenAI-driven fashion innovation.

Third, this study reveals the differentiated effects of various CPV dimensions in the context of GenAI fashion products. While perceived aesthetic value and perceived price value show significant impacts, perceived variety value does not significantly influence Chinese Millennials’ attitudes or purchase intentions toward CAGPs. This challenges the assumption that the diversity of GenAI-driven products automatically fulfills consumers’ needs for novelty and uniqueness. This finding provides empirical support for refining the applicability of the CPV framework in explaining consumer behavioral intentions toward GenAI fashion. Future research could further explore the potential moderating role of perceived authenticity or perceived uniqueness in the relationship between perceived variety value, attitude, and purchase intention.

5.2.2. Practical Implications

This study offers practical insights for developing and marketing GenAI technologies in premium clothing design by uncovering how Chinese Millennial consumers respond to key perceived values. First, the finding that the perceived design effort of premium fashion brands significantly influences Chinese Millennials’ attitudes and purchase intention. It suggests that these brands should clearly communicate the respective roles and contributions of human designers and GenAI in the co-creation process. Premium fashion brands should adopt new media marketing strategies—such as press releases and video demonstrations—to openly communicate with consumers, including Millennials. These efforts should highlight the role of human designers in generating creative ideas, ensuring design quality, and maintaining brand excellence. At the same time, brands should clarify that GenAI primarily serves as a supportive tool to help designers implement and adjust their ideas more efficiently. This type of messaging strengthens consumer confidence in the design quality of CAGPs launched by premium fashion brands. It also helps alleviate concerns that the use of GenAI may compromise design standards or diminish the unique creative style of human designers.

Second, perceived price value is a key determinant of purchase intention. Although the use of GenAI in the design and development process may reduce costs for premium fashion brands, they should avoid setting prices significantly lower than those of comparable existing collections. This helps maintain Millennials’ perceived value and demand for premium brands. At the same time, brands should not arbitrarily raise the price of CAGPs merely to align with fashion innovation or technological trends. Instead, pricing strategies should be based on the actual value that CAGPs deliver. Premium fashion brands should also transparently demonstrate to Chinese Millennials how they have used GenAI to improve design efficiency without compromising quality and how they leverage GenAI to analyze individual needs in depth to enhance personalized customization. In addition, brands may promote CAGPs on a limited basis through exclusive customization or limited editions. By increasing the perceived rarity and personalization of such products, Chinese Millennials are more likely to recognize the innovative and exclusive value of CAGPs, thereby accepting the pricing strategy and forming a more favorable perception of its price value.

Third, the significant effect of the subjective norm reflects the tendency of Chinese Millennials to place importance on social group evaluations. Given that this demographic tends to form product preferences through mobile media and online reviews, premium fashion brands should recognize the strategic value of key opinion leader (KOL) marketing. Brands can also collaborate with key opinion consumers (KOCs) by inviting them to experience CAGPs firsthand—such as wearing products or being shown the full design and development process. These immersive experiences can enhance their recognition of the product’s creative and personalized value and encourage them to share authentic feedback on their personal social media platforms for secondary dissemination. Compared to traditional advertising, content that showcases the real use of and personal experiences involving CAGPs is more likely to build trust and emotional resonance among Millennials. Positive social group opinions generated through such authentic sharing can reduce skepticism toward CAGPs and enhance the overall appeal of these products.

Finally, the aesthetic value of CAGPs remains critical to shaping Chinese Millennials’ attitudes. Premium fashion brands must leverage GenAI’s ability to autonomously learn fashion trends and consumer aesthetic preferences to design garments that not only align with current trends but also embody unique visual appeal, thereby capturing the attention of Millennials who value personalized self-expression. In addition, premium fashion brands should consider developing customized GenAI vision models that reflect their unique brand identity. Expert design aesthetics can be encoded into algorithms to train GenAI to generate fashion designs aligned with classical aesthetic principles. Human designers should remain responsible for curating creative inspirations and refining imperfect details throughout the design process. CAGPs that integrate trendy aesthetics with classical design principles are likely to enhance Millennials’ perception of aesthetic value. Altogether, these insights provide a comprehensive roadmap for premium fashion brands seeking to implement GenAI technologies in ways that align with consumer expectations and enhance competitive advantage in the digital fashion landscape.

6. Conclusions, Limitations, and Future Research

This study is the first empirical research on how Chinese Millennials perceive CAGPs for online shopping by premium brands and the factors influencing their purchase intention.

This research shows that perceived brand effort in design plays a key role in shaping Chinese Millennials’ attitudes and purchase intentions toward premium brands’ CAGPs, highlighting their focus on the brand’s design efforts. Perceived price value also significantly influences purchase intentions, emphasizing the link between pricing rationality and buying behavior. Perceived aesthetic values correlate positively with attitudes toward purchasing CAGPs, and subjective norms positively affect purchase intentions. While perceived variety value reflects Millennials’ pursuit of experiential value, its impact on attitudes and purchase intentions is limited. These findings suggest that Chinese Millennials prioritize a brand’s design effort, attitude, and image—particularly in terms of pricing—while valuing aesthetic value and social recognition of fashion trends. These conclusions enhance our understanding of Chinese Millennials’ decision-making across product, social, and personal dimensions.

Overall, this study offers valuable insights into Chinese Millennials’ feedback and consumption psychology regarding premium brands’ CAGPs for online shopping, and it also provides guidance to brands on their design, development, and marketing strategies for this new category.

This study provides an in-depth understanding of Chinese Millennials’ attitudes and purchase intentions toward CAGPs from premium fashion brands sold online. However, it also presents certain limitations that indicate directions for future research. The sample primarily consisted of participants from the Jiangsu–Zhejiang–Shanghai region recognized for their strong fashion industries and advanced e-commerce infrastructure. While this offers valuable insights into consumers with high exposure to fashion, it constrains the generalizability of the findings to Millennials from more diverse geographic, economic, and cultural backgrounds. Therefore, future research should expand the geographic scope of the sample. It should focus on Millennials in central and western regions, as well as those in third-tier and lower-tier cities, to enhance the generalizability of findings regarding attitudes toward CAGPs. Moreover, consumer attitudes toward AI-generated premium fashion may vary across cultural contexts, levels of digital literacy, and stages of market development. Cross-cultural studies are therefore encouraged to examine how differing cultural conditions shape perceptions of AI-driven fashion, contributing to a more comprehensive understanding of global consumer behavior. In addition, longitudinal research tracking the evolution of consumer attitudes and behaviors toward CAGPs may provide more nuanced insights into the long-term effects of AI-assisted fashion design.

The variables used in this study—such as perceived aesthetic value, attitude, and the subjective norm—are general in nature. However, to ensure a consistent and context-specific evaluation, all questionnaire items were worded to explicitly reference the AI context, and standardized visual stimuli were provided to guide respondents to focus specifically on CAGPs. However, future research could further develop and validate variables that more closely reflect the characteristics of CAGPs, such as perceived algorithmic creativity, perceived stylistic authenticity, and perceived sustainability. This would enhance the theoretical model’s explanatory power in the context of AI-generated fashion products. This study follows the TPB framework by using the subjective norm as a variable to measure social influence. However, future research could consider incorporating perceived social value as a complementary construct to more comprehensively explain symbolic and identity-driven motivations in AI-generated fashion consumption.

Given the quantitative focus of this study, future research could employ qualitative approaches—such as semi-structured or in-depth interviews—to gain deeper insights into consumers’ attitudes toward CAGPs. A multi-method approach that integrates both qualitative and quantitative methods is also recommended to develop a more comprehensive understanding of consumers’ attitudes and behavioral intentions related to CAGPs.