AI Technology Integrated Education Model for Empowering Fashion Design Ideation

Abstract

The rapidly increasing importance of technology integration and generative AI in the fashion industry is prompting changes in fashion design education. This study explored a new design methodology utilizing AI for sustainable and future-oriented fashion design education. The methodology involved consulting practitioners to select ChatGPT and Midjourney as AI tools and technology, pedagogy, and content knowledge (TPACK) as the theoretical framework. An AI prompt guide was developed based on specialized books, from which an educational program was created. An experiment with 30 third- and fourth-year fashion design students showed that their overall satisfaction with AI through TPACK was 4 out of 5, suggesting that TPACK enhances students’ creativity and efficiency through generative AI. Prompt guides received a satisfaction score of 4.7, indicating their usefulness for creative and efficient design outputs. AI-powered educational programs, like ChatGPT and Midjourney, also improved student creativity and learning efficiency, with ChatGPT scoring 4.5. However, concerns about technology dependency were noted. This study offers insights into integrating the latest technology into fashion design to improve process efficiency and creative output. This study not only provides a foundation for future research on AI design methodology but also explores practical directions for sustainable design in the fashion industry.

1. Introduction

The combination of advanced algorithms and human creativity in generative artificial intelligence (AI) technology is breaking new ground in various fields. Prompt engineering is at the core of this fusion. It consists of commands and parameters that direct the AI creation process. Prompt engineering facilitates the transformation of ideas into tangible outcomes and fosters creativity in generative AI. Platforms that offer generative AI provide examples of how to configure prompts, and users are creating prompting platforms that share prompts and images to make AI more accessible and usable. The market for generative AI is projected to experience rapid growth, reaching USD 1.452 trillion by 2030, a nearly 10-fold increase from USD 150.2 billion in 2023 [1]. Moreover, Pitchbook reported nearly USD 70 billion in investments in generative AI startups in 2023, with USD 29.1 billion allocated to generative AI, marking a growth of over 260% from the previous year [2].

Advances in generative AI have been shown to impact sustainability goals through design efficiency and innovation in the fashion industry [3,4]. LVMH, a prominent luxury brand group, announced the implementation of technology to create human-centered AI in partnership with Standford HAI, a human-centered AI initiative [5]. Additionally, PRADA, a subsidiary brand, featured fragrance campaign images produced using generative AI [6]. The interest in generative AI among various fashion brands, such as Collina Strada, Revolve, Gucci, Private Policy, Iris van Herpen, and Norma Kamali, continues to rise, expanding AI’s role from the fashion business to creative design [7]. Generative AI in the fashion industry has facilitated a sustainable fashion process by swiftly identifying trends, offering design concepts for new collections, streamlining the supplier supply chain, and reducing the time required for the design process. [8]. Consequently, fashion research utilizing generative AI has focused on trend prediction and style recommendations [9,10], the sustainable fashion revolution [11,12], customized textile design using AI technology [13], virtual fitting rooms and augmented reality [14,15,16], and AI–human collaborative design [17,18]. These developments underscore the pivotal role of AI in catalyzing innovation in the fashion industry, guiding it toward a more sustainable future, and emphasizing the need for specialized education to help designers comprehend and effectively utilize AI.

As generative AI technology progresses, the significance of prompt engineering is increasing. Works that visualize text prompts have gained recognition in the art world for their creativity, changing the way art is perceived. AI algorithms, when trained, can efficiently generate outcomes according to user input, sparking fresh ideas for designers. In the fashion industry, collaborating with designers through prompts to create collections is a game-changer for the design process. This approach could serve as an educational tool by fostering designers’ imagination and enabling them to explore new design concepts freely. Previous studies on prompt engineering have proposed theoretical frameworks for text-to-text generative AI, yet they tend to focus on various fields beyond design [19,20]. Marketing applications, particularly in the fashion industry, have been highlighted [21]. Regarding text-to-image generative AI, while some studies focus on modifying images using prompts [22,23] and assess the creativity of text-to-image generation [24], these works mostly address image alteration based on prompt composition rather than the design process itself. Considering this, limitations exist in applying these findings to practical design processes.

This study aims to develop an educational prompt guide for fashion design courses. Additionally, we seek to create an education program using generative AI based on the Technological Pedagogical Content Knowledge (TPACK) framework, a technology-integrated education model, implement it in a student setting, and conduct a survey to test its feasibility. We address the following research goals:

• To implement a technology-integrated educational model applicable to fashion design education;
• To develop a generative AI-based educational prompt guide for fashion design ideation;
• To evaluate the effectiveness of a design ideation educational program based on generative AI.

This study aims to provide a foundation for integrating generative AI into fashion design education, improving its quality. Additionally, we seek to explore new design methods that enhance students’ design thinking and creativity. The goal is to contribute to the fashion industry’s development by fostering a future-oriented educational environment.

2. Literature Review

This study systematically analyzes the experimentation of integrating generative AI into design education using cases of generative AI in art and design. We aim to understand the current state of the technology, highlight the importance of prompt engineering for design, and explore the effective use of AI tools. In addition, by reviewing technology education models, we propose an effective generative AI education methodology to lay a theoretical foundation for the integration of generative AI and fashion design education.

2.1. Generative AI in Art and Design

AI has revolutionized many fields by analyzing data, recognizing patterns, and learning through advanced technology [10]. It integrates various input methods, such as text and images, to help users communicate creators’ ideas [25]. Generative AI creates new data through deep learning based on input data called prompts using the AI’s generative model method. It generates sensible and ideal text, images, videos, speech, music, code, etc., that did not exist before [26,27,28]. Text-to-text-based generative AI used in various fields includes ChatGPT and BING, whereas text-to-image-based AI includes Midjourney, DALL-E, Stable Diffusion, and Adobe Firefly.

Generative AI has been particularly prominent in the arts, among other fields. In September 2022, Théâtre D’opéra Spatial, created with Midjourney, won first place in the digital art category of the Colorado State Fair’s annual art competition [29]. In March 2023, the Gagosian Gallery in New York City showcased a solo exhibition of AI artworks produced with DALL-E by filmmaker Bennett Miller, emphasizing the growing influence of generative AI in the art world [30]. This emergence of generative AI has also brought innovation to the fashion industry. AI applications expanding from online outfit customization services to stylist suggestions and design have broadened designers’ capabilities and enabled them to achieve diverse results [31]. Introducing the term “AI fashion designer,” the industry adopted AI Fashion Week in New York and Milan, merging fashion with avant-garde digital AI styles [32,33]. These changes signify more than just technological advancements; they are reshaping the traditional methods of design, production, and collection in the fashion world, creating new possibilities and avenues for creativity. Research on generative AI has shown the potential of collaborative design by proposing methods to increase design efficiency through frameworks, such as a sketch generation module and a rendering generation module, the latter of which learns the mapping between textures and sketches [34], as well as a human–AI collaborative design generation model to enhance creativity and social intelligence through collaboration between AI and human designers [18].

Generative AI is increasingly being used in fashion design to enhance designers’ creativity, innovation, and efficiency. AI can quickly suggest different designs through color, pattern, and silhouette variations, automating repetitive and technical parts of the design process. This has become an essential element and key strategy for staying competitive within changing fashion markets. Therefore, it is crucial that current and future designers are trained in generative AI integration to empower AI and lay the foundation for shaping the future of the fashion industry. Furthermore, the integration of technology would help designers quickly adapt to the latest trends and realize innovative ideas.

2.2. Prompt Engineering for Design

Prompt engineering [20,23,24,35,36,37], also known as prompting [37,38], prompt engineer [20], or prompt design [23,35], is a method for users to communicate with machine learning models by entering text [39]. In generative AI, prompt engineering involves selecting the appropriate words, keywords, and other essential steps to consistently generate the desired image [20,35,36]. This process is crucial for generative AI in text-to-text or text-to-image applications, as the quality of the prompt directly impacts the outcome [20]. As such, users must understand the language syntax and know how to modify prompts to achieve the desired results [37].

Prompt engineering in design involves converting text into images, with results varying based on prompt configuration. As the importance of prompt engineering has been highlighted, users should be able to adjust the text to refine the prompt for the desired outcome. Various prompt engineering approaches have been suggested as learning aids in design. One such approach is prompt engineer education [40,41,42], which is offered on image-generating AI platforms. It focuses on teaching users how to utilize the platform effectively and input optimized text. Next, the prompt guide [19,22,35,37,43,44] helps users understand the strategic construction of prompts, offering strategies and tips on organizing text to achieve the desired image. Additionally, prompting through experimentation by community members [45,46] aims to share real-world cases and creative efforts in formulating successful prompts. Finally, image-generative AI platforms, such as DALL-E, Midjourney, and Firefly, provide guidance on tool utilization and entering optimized prompts.

Prompt engineering provides a fast and effective means to the design domain by utilizing machine learning models. The different approaches allow for more fine-tuning and refinement during the iterative design process, and ultimately help produce a more polished design output. Nevertheless, prompt engineering is a counter-intuitive technique [37], and its practical application is still in infancy, with research on communication with AI required. There are currently a number of resources and guidelines available, but they are not yet comprehensive and in-depth. The effective application of prompt engineering requires detailed research into specific strategies and techniques that can effectively address different needs and creative challenges in design. Therefore, collaboration between various disciplines and continuous experimentation are essential for the continued development of prompt engineering to realize its full potential in the field of design.

2.3. Technology Education Model

The emergence of new digital technologies has revolutionized education beyond the advancement of technology [47,48]. Education no longer relies on traditional methods but adopts technology to change the way of learning. The integration of technology and education focuses on improving learners’ capabilities beyond simply mastering skills [49]. Teachers are exploring more creative and differentiated education methods. In this changing digital era, various education models have emerged as representative technology education integration models: substitution, augmentation, modification and redefinition (SMAR), technology integration matrix (TIM), technology integration planning (TIP), and TPACK. They aim to promote the integration of education and technology to provide students with a richer curriculum and learning experience.

Examining various technology integration models, the SMAR model was developed by Ruben Puentedura in 2010 to explore the effective use of technology in education through four stages of learning: substitution, augmentation, modification, and redefinition [50,51,52]. The TIM model, meanwhile, allows teachers to assess technology integration levels and set higher expectations using a rubric measuring five levels of integration and five learning environment attributes [53,54]. The TIP model, is a systematic method that includes planning, implementation, education, and evaluation phases focused on student outcomes and effectiveness [55]. The TPACK model underscores the importance of integrating pedagogy, content, and technology for effective curriculum design and implementation [56,57]. There is ongoing development and refinement of technology integration models by such organizations as the International Society for Technology in Education, including the ISTE Standards [58,59,60], UNESCO ICT Competency Framework for Teachers, a global guideline for developing teachers’ ability to integrate technology in education [61,62], and the European framework for the Digital Competence of Educators, a framework for European educators to develop digital competence and transform learning environments [49,63]. These efforts are contributing to the professionalization and quality of education by enhancing the digital competence of educators.

TPACK stands out among the various technology integration models for its ability to evaluate and apply educational technology from an integrated perspective. It provides specific guidance for deepening teachers’ understanding of technology and applying it in their educational practices. Essentially, TPACK focuses on planning and assessing teachers’ knowledge of how to effectively use technology in the classroom to enhance student learning experiences. This results in more meaningful, flexible, and effective knowledge and learning experiences for students [57].

TPACK’s framework originated with Shulman [64] and Gudmundsdottir and Shulman’s [65] pedagogy, content, and knowledge (PCK) models, and was later extended by Mishra and Koehler [56] and Koehler and Mishra [66]. TPACK is a progressive model consisting of seven fundamental elements of teacher knowledge: technological knowledge (TK), pedagogical knowledge (PK), content knowledge (CK), and their intersections technological pedagogical knowledge (TPK), technological content knowledge (TCK), pedagogical content knowledge (PCK), and TPACK (Figure 1).

TPACK’s effective, integrated approach to fashion design education contributes significantly to helping teachers develop strategies to foster students’ creative thinking. It can provide important guidance by introducing the latest technologies, such as generative AI, into the classroom. This integration of technology helps students keep pace with the latest industry trends, making the TPACK framework a suitable educational model for advancing fashion design education, including generative AI, and empowering students’ design competencies.

Overall, the integration of generative AI, prompt engineering, and technical education models in fashion design is opening up new possibilities for designers. These technologies enable efficient work while maintaining creativity and originality and promote innovation in the design process. However, this entails the need to understand and adapt to the technology. With continued study and improved education, these tools can play an important role in increasing creativity and efficiency in fashion design. In turn, the fashion industry would be able to produce more creative and efficient designs, and educational institutions would be able to effectively train future designers to become more competitive in the global market.

3. Methods

This study aims to explore the influence of fashion design inspiration and generate AI prompts using key design elements essential for design progress. The goal is to create and verify a novel education model for creative fashion design conception. We first develop an education model, then apply an education model, and finally we evaluate the education model based on prompts for design ideation (Figure 1).

3.1. Developing an Education Model

In the development phase of the education model, to adopt generative AI, we consulted five experts (see

Table 1. Fashion practice expert information.

	Job Title	Career	Field	Detail	AI Experience	AI Type
A	CEO	21 yrs	Women’s wear	Design, Pattern making	O	Chat GPT
B	CEO	9 yrs	Women’s casual	Design, 3D design	O	Chat GPT
C	CEO	8 yrs	Women’s wear	Design, Pattern making	O	Chat GPT, Midjourney
D	CEO	8 yrs	Women’s wear	Design, Marketing	O	Chat GPT, Midjourney, Stable Diffusion, Adobe Firefly
E	Team leader	6 yrs	Women’s wear	Design, Marketing	O	Chat GPT, Midjourney

) who have experience with or actively use generative AI in the fashion industry for advice on which generative AI programs have the most practical applications. These designers were brand owners, team leaders, or above, with hands-on insights and decision-making power, and their roles included design, pattern making, 3D design, and marketing. After consulting them and referring to articles on AI fashion [67,68,69], we selected ChatGPT and Midjourney as the research subjects, because they are currently used in the fashion industry and are suitable for cost-effective educational purposes. In addition, the experts suggested that the participant group be students who had already received creative design education in fashion design, as they would be able to make their own decisions and lead the design process without relying too much on AI when using generative AI. Thus, we recruited participants from the third and fourth years of fashion design majors who had already completed a basic fashion design course, and a total of 30 students participated in the experiment. Through this, we aimed to develop a technology education model and design an experiment to derive more accurate and effective results.

3.1.1. TPACK Design Ideation Education Model

To create an effective design ideation education model that utilizes technology, the TPACK framework was chosen. This model focuses on helping teachers plan and assess how to incorporate technology into their classes to enhance student learning. By applying the TPACK model, educators can integrate technology, content, and pedagogy to provide students with a richer learning experience and establish a new educational environment. This study determined that TPACK is suitable for developing and implementing an educational approach that introduces innovative technologies, like generative AI, into fashion design education. The seven core components of the TPACK framework as proposed by Koehler and Mishra [57] and Mishra and Koehler [56] were used to design ideation education.

Table 2. TPACK components and contents of fashion design ideation education using generative AI.

TPACK Component	Definition	Content
TK	An understanding of technology and the specific ways in which it is handled	Generative AI program types and leveraged knowledge
PK	Teacher’s knowledge of the process and practices or methods of learning	Curriculum and how it works
CK	Information about the teacher’s knowledge and the subject matter to be learned or taught	Fashion design education knowledge
TPK	Understand the types, components, and capabilities of educational technology and how it changes education and learning	Leveraging generative AI for fashion design ideation
TCK	Understand how technology and content impact and influence each other	Building the content of a fashion design prompt
PCK	Transforming curriculum for education	Educating fashion design ideation with generative AI
TPACK	A new type of knowledge that includes content, teaching methods, and technology	Fashion design ideation education model and outcomes enhanced through generative AI technology

summarizes the definitions of each component and how they can be integrated into design ideation education. The frameworks corresponding to TK, CK, and PK are described as different AI program types and leveraged knowledge in fashion design education, curriculum, and implementation. Next, TPK, PCK, and TCK, which correspond to the intersection of TK, CK, and PK, are applied in leveraging generative AI for fashion design ideation, building the content of a fashion design prompt, and educating fashion design ideation with generative AI. Finally, TPACK, which is the core of the entire education model, refers to the fashion design ideation education model and the education results using generative AI. Figure 2 visualizes the TPACK framework and components of fashion design ideation education using generative AI.

3.1.2. Developing a Guide for Design Ideation Prompts

The prompt guide was developed by consulting fashion design books and various studies on AI prompts to enhance ChatGPT and Midjourney for fashion design education. The guide, based on Seivewright and Sorger’s [70] research and design for fashion, is structured in two phases: the design inspiration phase, in which students find inspiration and select a topic through research; and the design idea generation phase, in which students develop design ideas by integrating fashion design elements. “Design development” and “product development”, cited in Fashion: From Concept to Consumer [71,72], were referenced to form the basic framework of the fashion design ideation prompt guide with generative AI. Fashion Design Research by Mbonu [73] aided in reviewing the concept board in the design inspiration phase and the design ideation board in the design idea generation phase. Resources related to AI prompts were drawn from previous studies that detailed prompt construction [19,20,22,35,37], professional books [74,75,76], and official materials from Midjourney [41], an image-generating AI platform utilized in the research.

The initial design ideation prompt guide, developed from this process, was pre-tested with five current fashion industry designer experts (Figure 1). Following the test, in-depth interviews were conducted with the designers to assess their satisfaction and evaluate its potential for use in education and practice. Specific examples were then added to each step of the prompt guide based on their feedback to ensure its effectiveness in real-world education and practice. Additionally, a new step was included to support the creation of prompts that can generate diverse images, which involved using ChatGPT to categorize and organize prompts related to a theme or idea into various adjectives, nouns, and verbs.

The fashion design ideation prompt guide with AI overview is divided into two phases: design inspiration phase and design idea generation phase. The design inspiration phase includes “inspiration research,” “define design concept”, “build prompts”, “exact images”, “modify and complement prompts”, and “concept board”. The design idea generation phase includes “design element research”, “create design ideas”, “build prompts”, “exact images”, “modify and complement prompts”, and “design ideation board” (Figure 3).

Design Inspiration Phase

In the prompt guide, the design inspiration phase consists of seven steps, with the first three steps utilizing ChatGPT and the last four steps utilizing Midjourney. In the first step, a topic area is set and various resources collected through ChatGPT to define a specific design topic. In the second and third steps, a set of prompts is built based on the collected data, which are organized into a syntax. The fourth step is to organize the content of the prompts to be entered into Midjourney, and the fifth step is to extract multiple images from Midjourney and select the desired ones. The sixth step is to reorganize the images by modifying and complementing the prompts based on the selected images. The last step is to extract the final image with the revised prompts and to finalize the concept board (refer to Appendix A for details).

Design Idea Generation Phase

In the prompt guide, the design idea generation phase consists of six steps, utilizing ChatGPT for the first two steps and Midjourney for the last four steps. The first step utilizes ChatGPT to build design idea prompts based on fashion design elements. In the second step, the prompts are organized into a syntax to build a variety of design ideas. In the third step, the organized prompts are combined to develop specific design ideas. In the fourth step, Midjourney is used to extract a variety of design images from the prompts built. In the fifth step, the prompts are modified and complemented based on the extracted images. The last step is to create a final design ideation board with the revised prompts (refer to Appendix A for details).

3.2. Applying an Education Model

3.2.1. Recruitment and Selection of Research Participants

In the stage of applying an education model, 30 students in the third and fourth year of fashion design majors, who had completed basic to advanced courses related to fashion design and had experience in portfolio production, were selected as empirical research participants. The study participants were chosen based on recommendations from five current design experts in the fashion industry and two experts in fashion design education. These experts pre-tested the fashion design ideation education program utilizing generative AI. The participants were students capable of determining the final goal image for their design ideas through their own projects, and had basic digital literacy skills that facilitated their understanding and use of generative AI as a design tool. The selection criteria allowed us to compare the traditional design education experience with that of design education utilizing AI.

3.2.2. Education Program with Generative AI Prompt Guide

In this study, a three-session education program was designed to introduce students to how to use generative AI in the design ideation stage of fashion design development. The program included 30 students divided into two groups of 15, with morning and afternoon sessions available. Each session lasted 60 min, with 50 min of class time and a 10-min break. To enhance the program’s quality and maximize learning, each session was two faculty members and one teaching assistant.

The first session focused on TPACK’s TK, PK, and CK to teach generative AI and fashion design ideation. Students learned the basic principles of ChatGPT and were introduced to the Midjourney platform so that they could understand and familiarize themselves with its usage. They also comprehended and explored how the basic elements of fashion design could be utilized in generative AI by understanding the basic content and composition of fashion design. Students’ questions and concerns were resolved through Q&A. In preparation for the second session, five teams of three students collaborated to install AI programs, research design ideas, and create prompts for problem-solving. This collaboration aimed to increase efficiency by diversifying ideas and maximizing creativity in the problem-solving process. The program’s theme was “subcultures”, providing a basis for comparing traditional and AI-assisted design ideation methods to understand their differences.

The second session focused on TPACK’s TPK, PCK, and TCK, with the objective of design inspiration. The students were divided into teams and followed the “design inspiration phase” sequence provided in the prompt guide. Each team utilized ChatGPT to research ideas, choose a design topic, and set up AI prompts to generate images related to the topic. The teams then entered the prompts into Midjourney to produce concept boards for the design topic.

The aim of the third session was to create design ideas based on TPACK’s TPK, PCK, TCK and integrated knowledge. Students followed the “design idea generation phase” sequence in the prompt guide. Each team used ChatGPT to explore elements of fashion design, such as color, fabric, shape and structure, details, and print and surface decoration. They developed design prompts and fed them into the image-generating AI to complete the design ideation board. The finished design ideation boards, along with the previous concept boards, were presented to the teams for feedback. The final stage of the educational program involved conducting a survey and evaluation of the participating students to collect feedback on the program using the generative AI prompt guide.

Table 3. Fashion design ideation education program using generative AI.

Program Plan				TPACK Competency
1st session	Session objective		An introduction to generative AI and fashion design ideation education	TK PK CK
	Session content	1.	Utilizing generative AI (Chat GPT, Midjourney) Understanding the basic elements that comprise fashion design
		2.	Understanding the basic elements that comprise fashion design
	Key points		Organizing team, introducing main topic
2nd session	Session objective		Design inspiration	TPK PCK TCK
	Session content	1.	Design idea research, team topic selection, and prompt organization with a generative AI Chat GPT
		2.	Creating concept board with image-generating AI Midjourney
	Key points		Working in teams, using the prompt guide “Design inspiration phase,” Creating concept board
3rd session	Session objective		Design idea generation	TPK PCK TCK TPACK
	Session content	1.	Building design prompts based on fashion design elements with Generative AI, Chat GPT
		2.	Creating sketches of design ideas with image-generative AI Midjourney
	Key points		Working in teams to generate design ideas using the prompt guide “Design Ideation Phase”. Creating a design ideation board, Presenting and submitting the results, Surveying and evaluating the education program.

summarizes the content of the fashion design ideation education program using generative AI.

3.3. Evaluating the Education Model

This study was approved by the researchers’ Institutional Review Board (IRB No. 1044396-202310-HR-215-01). The evaluation of the education model was carried out using a semi-structured questionnaire that included a 5-point Likert scale, multiple-choice questions, importance ranking, and open-ended questions. The questionnaire was divided into two parts: a preliminary survey to assess students’ prior knowledge and experience in a fashion design education program utilizing generative AI, and a post-survey based on the TPACK, which includes TPK, TCK, PCK, and TPACK. The TPACK-based questionnaire aimed to comprehensively evaluate the impact of the technology-enhanced education program on students’ technical, content, and pedagogical knowledge and skills. It was used to gain a deeper understanding of the study subject and collect a wide range of information and data from the participants.

For the questionnaire, the initial questions were created to explore students’ basic awareness and experience with generative AI. The aim was to determine the level of knowledge students had about generative AI technologies, their previous use of AI tools, the reasons for their experiences, and their proficiency in understanding and applying AI technologies effectively. We aimed to investigate whether technical background and practical experience were linked to adaptability and creative application during the learning process.

Moving on to the post-questionnaire based on TPACK, the TPK section included questions centered on the use of AI for fashion design ideation. Split into the design inspiration phase and design idea generation phase, we gathered feedback on whether utilizing generative AI tools aided in design ideation and the differences and enhancements between traditional design ideation and generative AI-based ideation. Through this research, we aimed to examine how generative AI could be integrated into fashion design education and gain a deeper understanding of the impact of AI technology on learners’ creative process and ideation methods.

The TCK section comprised questions focusing on the efficiency of the fashion design prompt content developed for AI usage. Segmented into the design inspiration phase and design idea generation phase, this section collected opinions on whether the prompt guide provided in the educational program was beneficial for using generative AI and if participants would consider using the prompt guide in the future with AI. By assessing users’ experiences, our goal was to evaluate the usefulness of prompt guides for AI utilization and identify ways to enhance and refine the prompt guide in the future.

The PCK section included questions about the effectiveness of AI in fashion design ideation education. The section was divided into the design inspiration phase and the design idea generation phase to gather opinions on the pros and cons of utilizing AI in fashion design education, as well as specific feedback on AI’s assistance. We also asked about satisfaction with AI in design education and how its quality compared to traditional methods. Our goal was to understand the impact and effectiveness of design ideation education using AI when it came to students’ design outcomes.

The questions in the TPACK section focused on the overall evaluation of and satisfaction with the fashion design ideation education program using generative AI. Based on the education program, participants were asked to evaluate their satisfaction with the use of generative AI technology in fashion design ideation education, and to provide insight into the possibility and rationale for recommending the use of generative AI in future fashion design classes. Suggestions were also gathered on the stages at which generative AI could be utilized in the fashion design education process, and practical data were collected on effectively integrating AI technology into fashion design education. Additionally, a satisfaction survey was conducted to assess the quality and effectiveness of the program. The feedback obtained was intended for use in improving the curriculum of fashion design education and exploring better ways to integrate AI technology.

4. Results

In this study, 30 students who participated in a design ideation education program using generative AI participated in a creative design process using a prompt guide. They explored new approaches to fashion design using generative AI technology, following which they completed a survey based on the TPACK education model. Based on these survey results, we analyzed the effectiveness of generative AI in education.

4.1. Preliminary Survey

To optimize the education program for students’ needs and levels, a survey was conducted on learners’ awareness of generative AI, frequency of use, experience with digital tools, and purpose of use. The survey utilized a 5-point Likert scale, multiple choice questions, and open-ended questions (

Table 4. Preliminary questionnaire (5-point Likert scale).

Pre-question	Questions		Mean
	#1	Do you know about generative AI?	3.0
	#2	How often do you use generative AI?	1.7 * F-Likert
	#3	How would you rate your digital literacy skills in design?	3.4

1: Poor; 2: Fair; 3: Good; 4. Very good; 5: Excellent. * F-Likert: Frequency-based Likert scale, 1: Very rarely; 2: Rarely; 3: Occasionally; 4. Frequently; 5: Very frequently.

,

Table 5. Preliminary questionnaire (multiple choice questions).

Pre-question	Questions
	#1	Which generative AI have you used? (You can select multiple) Options: Text generative AI (Chat GPT), Image generative AI (DALL-E, Stable Diffusion, etc.), Music generative AI, AI translation tool, Speech generation and conversion AI
	#2	Which design tools have you used? You may select multiple options: Options: Adobe Photoshop, Adobe Illustrator, CLO 3D, digital drawing and illustration app (Procreate, other similar apps), CAD/3D modeling tool (Auto CAD, SketchUp), InDesign

and

Table 6. Preliminary questionnaire (open-ended questions).

Pre-question	Questions
	#1	What is the name of the image-generating AI you have used?
	#2	What have you mainly used generative AI for?

). The results revealed that learners’ digital literacy was above average at 3.4, with average awareness of generative AI at 3. However, their frequency of using it was low at 1.7, indicating limited experience with generative AI. The types of generative AI utilized included text generative AI (ChatGPT), AI translation tools, image generative AI (Midjourney, DALL-E), and voice generation and conversion AI. Concerning image generative AI, most respondents had not used any (83.3%), followed by Midjourney (14.3%) and DALL-E (3.6%). The main reasons for using generative AI were for assignments (37%), to aid in learning (29.6%), and for personal hobbies or creative pursuits (25.9%). As for digital design tools, Adobe Illustrator, Adobe Photoshop, CLO 3D, digital drawing and illustration apps (e.g., Procreate), CAD/3D modeling tools (e.g., Auto CAD, SketchUp), and InDesign were commonly used, with Adobe Illustrator (100%) and Adobe Photoshop (96.7%) being the most popular. This demonstrates that most students had a good grasp of digital tools for fashion design ideation.

4.2. TPK Competency

The survey examining the use of generative AI in fashion design ideation corresponding to TPACK’s TPK was divided into two phases: design inspiration and design idea generation. It was conducted with a 5-point Likert scale and open-ended questions (refer to

Table 7. TPK: Leveraging generative AI for fashion design ideation (5-point Likert scale).

TPK	Phase	Questions		Mean
	Design inspiration phase	#1	Did you find the utilization of generative AI (Chat GPT, Midjourney) helpful during design inspiration phase?	4.6
		#2	Do you think that utilizing generative AI (Chat GPT, Midjourney) for design inspiration is more effective than traditional design inspiration methods?	3.9
	Design idea generation phase	#3	Did you find the utilization of generative AI (Chat GPT, Midjourney) helpful during the design idea generation phase?	4.6
		#4	Do you believe that using image-generating AI (Midjourney) during the design idea generation phase results in better design ideas compared to traditional methods?	3.8

1: Poor; 2: Fair; 3: Good; 4. Very good; 5: Excellent.

and

Table 8. TPK: Leveraging generative AI for fashion design ideation (open-ended questions).

TPK	Phase	Questions
	Design inspiration phase	#1	What improvements could be made to utilize generative AI (Chat GPT, Midjourney) in design inspiration phase?
	Design idea generation phase	#2	How can generative AI (such as Chat GPT and Midjourney) be better utilized in the design idea generation phase?

). Both the design inspiration and design idea generation phases received high ratings of 4.6 in terms of the helpfulness of using generative AI for fashion design ideation. Additionally, the comparison between generative AI and traditional design ideation methods received positive ratings of 3.9 in the design inspiration phase and 3.8 in the design idea generation phase. This indicates that incorporating new generative AI technologies in design ideation has been beneficial for developing design ideas. However, feedback suggested that designers face challenges expressing emotions and conveying meaning accurately through AI. Improving the user interface and the ability to produce diverse results were highlighted as areas for enhancing AI usability and the quality of creative outcomes.

Design inspiration phase

“It would be great if AI could understand keywords in the emotional area of fashion.

“I wish generative AI could understand the meaning of prompts more accurately.

“It would be nice to be able to manipulate the details of an image directly instead of through text.

“Also, it would be nice to see more examples than just four.”

Design idea generation phase

“The AI doesn’t seem to understand a lot of fashion design jargon.”

“It would be nice to have a step before the prompt to avoid distorting the designer’s intentions.”

“I feel like this is a limitation of the data and algorithms and would like to see this expanded.”

4.3. TCK Competency

The survey on the construction of fashion design prompt contents corresponding to TPACK’s TCK was divided into two phases: design inspiration and design idea generation. Participants rated their satisfaction using a 5-point Likert scale (

Table 9. TCK: Building the content of fashion design prompts (5-point Likert scale).

TCK	Phase	Questions		Mean
	Design inspiration phase	#1	Was it helpful to follow the prompt guide presented in the design inspiration phase to build prompt content in generative AI (Chat GPT)?	4.7
		#2	Did the prompts developed during design inspiration phase aid in the utilization of image-generating AI (Midjourney) for design ideation?	4.7
		#3	Do you plan to continue utilizing the design prompt guide for generative AI (Chat GPT, Midjourney) in the future design inspiration phase?	4.4
	Design idea generation phase	#4	Was it helpful to follow the prompt guide presented in the design idea generation phase to build prompt content in generative AI (Chat GPT)?	4.7
		#5	Did the prompts developed during design idea generation phase aid in the utilization of image-generating AI (Midjourney) for design ideation?	4.6
		#6	Are you planning to continue using the design prompt guide for generative AI (Chat GPT, Midjourney) in future design idea generation phases?	4.3

1: Poor; 2: Fair; 3: Good; 4. Very good; 5: Excellent.

). The results indicated consistent high satisfaction levels of 4 points or more throughout the design ideation phase. In particular, participants rated the helpfulness of building prompt contents based on the guide provided in ChatGPT highly, with both the design inspiration and design idea generation phases receiving a rating of 4.7, signifying very high satisfaction. Midjourney, an image-generating AI, also received high ratings for prompt construction, scoring 4.7 for the design inspiration phase and 4.6 for the design idea generation phase. Participants expressed positive intentions to use the prompt guide in the future, with a rating of 4.4 for the design inspiration phase and 4.3 for the design idea generation phase.

4.4. PCK Competency

The survey on fashion design ideation education using generative AI based on TPACK’s PCK was conducted in two phases: design inspiration and design idea generation. It included 5-point Likert scale and multiple-choice questions (

Table 10. PCK: Educating fashion design ideation with generative AI (5-point Likert scale).

PCK	Phase	Questions		Mean
	Design inspiration phase	#1	Are you satisfied with the concept boards visualized through the fashion design ideation education utilizing generative AI in design inspiration phase?	4.1
		#2	Do you think the quality of the completed concept boards, generated through fashion design ideation education utilizing generative AI, is better than those completed through traditional design inspiration methods?	3.9
	Design idea generation phase	#3	Are you satisfied with the results of the design ideation boards visualized through fashion design ideation education utilizing AI in the design idea generation phase?	4.0
		#4	Do you think the quality of design ideation boards created through fashion design education using generative AI is better than those made with traditional design methods?	3.9

1: Poor; 2: Fair; 3: Good; 4. Very good; 5: Excellent.

and

Table 11. PCK: Educating fashion design ideation with generative AI (multiple choice questions).

PCK	Phase	Questions
	Design inspiration phase	#1	If fashion design ideation education using generative AI (Chat GPT, Midjourney) was helpful during the design inspiration phase, which aspects were beneficial? (Multiple selections possible, specify “none” if not applicable) Options: Topic and concept development, Visual idea generation, Promotion of creativity and originality, Time and resource saving
		#2	When compared to traditional design inspiration methods, what are the advantages of using generative AI (Chat GPT, Midjourney) for fashion design ideation education? (Multiple selections possible, specify “none” if not applicable) Options: Improvement in speed and efficiency, Promotion of creative ideas, Diversity of information, Rapid problem-solving
			When compared to traditional design inspiration methods, what are the drawbacks of using generative AI (Chat GPT, Midjourney) for fashion design ideation education? (Multiple selections possible, specify “none” if not applicable) Options: Limitations in in-depth analysis, Limitations in providing customized solutions, Issues with data accuracy and reliability, Limits in research creativity, None
	Design idea generation phase	#3	If fashion design ideation education using generative AI (Chat GPT, Midjourney) was helpful during the design idea generation phase, which aspects were beneficial? (Multiple selections possible) Options: Development of styles and silhouettes, Combination of colors, Generation of materials and patterns, Easy design modification and iteration, Creativity in design, Various design options
		#4	When compared to traditional design idea generation methods, what are the advantages of using generative AI (Chat GPT, Midjourney) for fashion design ideation education? (Multiple selections possible) Options: Improvement in speed and efficiency, Creative inspiration, Infinite possibilities for variation, Ease of modification, Effective simulation of design ideation
			When compared to traditional design idea generation methods, what are the disadvantages of using generative AI (Chat GPT, Midjourney) for fashion design ideation education? (Multiple selections possible) Options: Limitation in expressing designer sensibility, Constraints on design creativity due to limitations in data and algorithms, Limitation in perfect control over the output, Misinterpretation of design intentions owing to communication issues with AI, Occurrence of technical dependency

). The satisfaction rate for the concept board in the design inspiration phase was 4.1, while for the design ideation board in the design idea generation phase it was 4.0, both considered very high. Comparing the quality of results with traditional design ideation methods, AI-based design ideation education received a score of 3.9, above average.

The primary reasons for the high satisfaction with fashion design ideation education using generative AI included visual idea generation (83.3%) and concept development (80%) in the design inspiration phase, and design creativity (70%), style and silhouette development (66.7%), and design options (50%) in the design idea generation phase. The advantages of fashion design ideation education using AI included improved speed and efficiency (86.7%), creative idea promotion (73.3%), and information diversity (60%). However, data accuracy and reliability (56.7%) and limitations of customizable solutions (53.3%) were mentioned as disadvantages.

In the design idea generation phase, the highest-ranking results were improvement in speed and efficiency (90%), creative inspiration (66.7%), effective simulation of design ideation (63.3%), and infinite variation possibilities (53.3%). However, users cited misunderstandings of design intent owing to communication issues with the AI (66.7%), limited designer emotional expression (60%), and a lack of complete control over the results (60%).

4.5. TPACK Competency

The survey on the education model and results of fashion design ideation through generative AI technology, based on the TPACK framework, consisted of a 5-point Likert scale, multiple-choice questions, and rankings (

Table 12. TPACK: Fashion design ideation education model and education results using generative AI (5-point Likert scale).

TPACK	Phase	Questions		Mean
	Education program	#1	Do you think it would be beneficial to use generative AI (Chat GPT) technology in fashion design ideation education?	4.5
		#2	Do you think it would be beneficial to use image generative AI (Midjourney) technology in fashion design ideation education?	4.5
		#3	Would you recommend using generative AI (Chat GPT) in future fashion design courses?	4.2
		#4	Would you recommend using image-generative AI (Midjourney) for future fashion design courses?	4.1
		#5	Are you satisfied with the design ideation education program process utilizing generative AI?	4.4

1: Poor; 2: Fair; 3: Good; 4. Very good; 5: Excellent.

,

Table 13. TPACK: Fashion design ideation education model and education results using generative AI (multiple choice questions).

TPACK	Phase	Questions
	Education program	#1	If you were to recommend the use of generative AI (Chat GPT) in future fashion design courses, what would be the reasons? (Multiple selections possible) Options: Providing diverse information, Generating design ideas, Saving time and convenience, Efficient collaboration tool, Individualized tailored feedback
			If you were not to recommend the use of generative AI (Chat GPT) in future fashion design courses, what would be the reasons? (Multiple selections possible) Options: Lack of reliability of information, Limitation of creativity, Increased dependency on technology, Deterioration of the essence of learning process, Creation of learning gaps based on individual accessibility and utilization capabilities
		#2	If you were to recommend the use of image-generating AI (Midjourney) in future fashion design courses, what would be the reasons? (Multiple selections possible) Options: Providing rapid visual feedback, Enhancing diversity in design ideas, Increasing efficiency in the design process, Serving as an efficient collaboration tool, Enhancing expressive capabilities through technology
			If you would not recommend the use of image-generating AI (Midjourney) in future fashion design courses, what reasons would you choose? (Multiple selections possible) Options: Concerns about creative deterioration, Unreliable technology reliability, Deviation from the essence of the learning process, Excessive dependency on technology, Issues with accessibility and equity in technology utilization

and

Table 14. TPACK: Fashion design ideation education model and education results using generative AI (ranking).

TPACK	Phase	Questions
	Education program	#1	If generative AI tools like Chat GPT and Midjourney were to be implemented in future fashion design education programs, in which stage of the design process do you believe they should be utilized? Please select the following options in order of importance: Options: Design ideation (Research, Concept board), Design ideation (Design idea generation, Design ideation board), Prototype development (Collection development, finished designs), Create lookbooks and Photo shoots, Create an overall portfolio

). The results of the survey showed that the use of ChatGPT and Midjourney was perceived as very effective, with a high rating of 4.5. The use of generative AI tools in future fashion design classes was perceived positively, with ratings of 4.2 for ChatGPT and 4.1 for Midjourney. The main reasons for recommending ChatGPT were information variety (73.3%), time-saving convenience (66.7%), and design idea generation (63.3%). The main reasons for recommending Midjourney were increased variety in design ideas (73.3%), quick visual feedback (70%), and efficiency in the design process (60%). Conversely, reasons for not recommending ChatGPT included over-reliance on technology (75%), detracting from the essence of learning (42.9%), and information reliability concerns (42.9%). Reasons for not recommending Midjourney included excessive dependence on technology (78.6%) and fears of stifling creativity (57.1%).

In the future, when integrating generative AI into fashion design education, the most suitable design stages to utilize are: first, the design inspiration stage of design ideation; second, the design idea generation stage of design ideation; third, the prototype development stage of collection development and completed designs; fourth, the lookbook and photo shooting stages; and fifth, the overall portfolio creation stage. The design ideation education program using generative AI was evaluated with a high satisfaction score of 4.4.

5. Discussion

5.1. Need for a Technology-Integrated Fashion Design Educational Model

This study was conducted based on the TPACK education model to implement a technology-integrated education model that can apply generative AI to fashion design education.

The study focused on the initial stage of the design process, design ideation, and developed a guide and education program for the use of generative AI, ChatGPT and Midjourney, in the design inspiration phase and design idea generation phase and evaluated their effectiveness. As a result, the average score was 4.2 for TPK (leveraging generative AI for fashion design ideation), 4.5 for TCK (building the content of fashion design prompt), 3.9 for PCK (educating fashion design ideation with generative AI), and 4.3 for TPACK (fashion design ideation education model and education results using generative AI). In other words, the technology-integrated fashion design education increased creativity and efficiency by allowing students to quickly explore and visualize various and complex design options, unlike traditional design ideation methods. This was consistent with the experimental results of Yan et al., who found that designing with generative AI had a positive effect on creativity and efficiency [34].

As new technologies like generative AI continue to evolve rapidly, education models must continue to develop alongside them. In this way, educational programs that reflect the latest technologies could help students effectively adapt to the future industrial environment. The high level of human-centered control acquired in this way, along with high levels of computer automation, will further enhance human performance and make it more widely available [77]. Such technology integration in fashion design education will make the design process more efficient and flexible, which in turn will improve the quality of design and expand the range of creative outputs. This will play an important role in increasing design competencies and revolutionize both design education and practice.

5.2. Importance of Developing Educational AI Prompts Optimized for Fashion Design Ideation

In generative AI, it is important that prompts accurately reflect the user’s intentions and needs to achieve the desired outcome. In this study, we aimed to develop an educational fashion design ideation prompt guide using generative AI.

The results of the study showed that the use of ChatGPT and Midjourney as the prompt guide in the design inspiration phase received a score of 4.7, and in the design idea generation phase, ChatGPT received a high score of 4.7 and Midjourney received a high score of 4.6, suggesting that they were helpful in building the content of generative AI prompts. Additionally, when asked if they would use the prompt guide in the future, ChatGPT received a high score of 4.4 in the design inspiration phase and 4.3 in the design idea generation phase. This proves that the prompt content construction based on the provided prompt guide effectively supports the use of generative AI in the design ideation process. Students were highly satisfied with using the prompt guide to build their prompts and were positive about using it in the future. In this way, the prompt guide helped them to utilize the capabilities of generative AI in a deeper way, which in turn helped them to achieve more creative and effective results. These findings are consistent with the results of Korzynski et al. [19] that prompt engineering enables new forms of creative expression in the arts, and that effectively crafted prompts can improve outcomes. Furthermore, Oppenlaender et al. establish the importance of developing discipline-specific prompts [37], which suggests that prompt creation and refinement through rich descriptive language can change the quality of a piece of work. However, their study questions whether it can be utilized intuitively without specialized training and understanding, and discusses the need for deep subject-matter expertise to effectively utilize generative AI [37].

Related to this is the current lack of prompt guide specifically tailored to address the inspiration and design elements unique to the fashion design process. Existing guides designed for general purposes, such as illustration and animation [40,41,42,43,44,45,46], have proven inadequate for the complexity of fashion design, highlighting the need to develop specialized prompt guides specific to fashion design. The development of user-centered prompt guides is an important step in enabling generative AI to produce more thoughtful and creative outputs. Through prompts, users should be able to communicate their creative vision and specific requirements to the AI, including precise word choice and specific and detailed instructional contexts [43,44]. In other words, designers and students need to learn how to articulate their own design thinking by understanding fashion design elements and processes, and acquire the skills to write professional prompts based on this knowledge. This would enable prompts to be used as a material for new designs, supporting new design methods and providing a future for creative work [27], and would develop the ability to effectively realize ideas.

5.3. Educational Value of Fashion Design Ideation Based on Generative AI

The advent of generative AI has revolutionized the field of design, and the effective utilization of these technologies requires appropriate pre-education. In this study, we evaluated the usefulness of design ideation education programs based on generative AI.

The results of the study showed a high level of satisfaction with the use of ChatGPT and Midjourney in a design ideation education program based on the TPACK framework, with a score of 4.5 for ChatGPT and 4.1 for Midjourney, and a very high level of satisfaction with the process of the design ideation education program using generative AI, with a score of 4.4 for ChatGPT and 4.2 for Midjourney. Students were highly satisfied with the variety of information they received in ChatGPT (73.3%) and the diversity of design ideas they generated in Midjourney (73.3%). At the same time, however, students were concerned about the increased reliance on technology in ChatGPT (75%) and the excessive reliance on technology in Midjourney (78.6%). These dual findings, especially given that the students in the study were upper-level students with experience creating portfolios, point to the importance of educating students to understand and effectively apply generative AI technologies beyond just using them as tools. This means that it is essential to teach students to use AI tools ethically [78], and to empower learners and foster critical thinking skills through continuous learning with new technologies [79].

The results suggest that design education with generative AI should address both technical and design aspects. In terms of technical education, students should be taught the types of generative AI programs and how to use them, how to use generative AI for fashion design ideation, and how to build the content of fashion design prompts to help them develop the ability to solve problems that arise from interacting with machines. This knowledge will enable learners to actively use the technology in a variety of areas, from how to collaborate with AI, to developing prompts and interpreting results. In terms of design education, learners should also be taught how to use AI as an aid to creative design, visualizing ideas and using critical thinking and analytical judgment to effectively use it throughout the design process. In this way, it is important for learners to develop the ability to make choices among the various ideas and design options presented by AI through education programs that best meet their intentions and goals. In the end, a training program that combines these two pedagogical approaches would set learners up for more innovative and creative work in collaboration with AI.

6. Conclusions

This study verified the importance of prompts that facilitate the development of generative AI and applied them to the fashion design process. To this end, we implemented the TPACK technology integration education model, developed a prompt guide for fashion design ideation using generative AI, applied a design ideation education program based on generative AI to students, and evaluated its usefulness. Through this process, we proposed and tested a new design idea enhancement method using generative AI in fashion design, with the following implications.

First, this study makes a novel academic contribution by proposing a new design method applying the latest technology to the field of fashion design. By demonstrating the effectiveness of a fashion design education model utilizing generative AI, this study provides a new paradigm for fashion design education, which is an important resource for the development of fashion design theory. In addition, this study confirmed that the framework for utilizing generative AI based on the TPACK education model is suitable for fashion design education, laying the foundation for further academic exploration of the latest technologies.

Second, this study has educational implications by developing and evaluating an educational program that effectively integrates generative AI in fashion design education. In terms of education, this study proposed a practical teaching method to apply creative thinking and innovative design skills through a fashion design generative AI prompt guide based on the TPACK model. Through the design ideation education program using generative AI, educators were able to enhance students’ design creativity and efficiency. In addition, the development and utilization of prompt guides helped students interact with AI to produce better design outcomes. This pedagogical approach can help students adapt to the future industry environment and cultivate the skills needed to actively utilize new technologies.

Third, the fashion design education model utilizing generative AI has important practical implications. In the fashion design industry, creativity and efficiency are essential for competitiveness. In this study, practicing designers were consulted in the process of creating prompts for generative AI, and the prompt guide was revised to reflect their perspectives. This study demonstrates that generative AI can help designers explore and visualize different design options more quickly and with more variety. This can increase the efficiency of the fashion design process and improve the quality of creative output. In addition, it is believed that those educated in design using generative AI would be able to effectively utilize the new technology to provide innovative design solutions, enabling generative AI to make significant contributions to the industry beyond fashion design education.

The main limitations of this study are its application of generative AI to the design ideation stage of the fashion design process as well as the small number of students who participated in the research experiment. Because the experiment was conducted only in the design ideation stage, it was not possible to evaluate the effectiveness of generative AI in other stages of the design process. Therefore, future research should conduct a comprehensive verification and evaluation of how generative AI can be utilized in various processes, such as prototype development (collection development, finished design), creating lookbooks and photoshoots, and creating overall portfolios, in addition to design ideation. The number of students participating in the study was limited to 30, which limits generalizing the results. Nevertheless, this study supplemented our quantitative survey results with qualitative content by continuously and closely observing the fashion design students’ first experiences and practice with the AI design prompt guide and conducting in-depth interviews with the students.

In future research, we aim to divide the fashion design process into stages, provide prompt guides for each stage, and experiment with learning materials specific to each stage. We also seek to increase the sample size to include students from different backgrounds and grade levels. This would allow us to draw more comprehensive and reliable conclusions about the use of generative AI in fashion design education. Finally, we aim to develop generative AI prompts that are practically applicable to designers and aspiring design entrepreneurs so that they can utilize them as a design development tool to meet the needs of the fashion industry.