Navigating the Confluence: Advancements and Trends in Artificial Intelligence-Driven Design Education Based on CiteSpace's Visual Analytics

Abstract. The arrival of the artificial intelligence era has had a huge impact on the design industry. Design education needs to transform to meet the new demands of the 21st century. This study aims to explore the intersection and integration of artificial intelligence and design education through scientometric analysis and to construct a visual knowledge map. Firstly, utilizing the scientific tool CiteSpace, a visual map is constructed based on data from 810 articles in the Web of Science core collection database. Furthermore, the map includes analyses of national collaborations, institutional research areas and classifications, co-occurrence of keywords, and cluster analysis. Secondly, the generated visual map is analyzed to understand the current research status, hot topics, and future trends. The research findings fill the existing literature gap in the combination of design education and artificial intelligence, providing stakeholders with timely and effective access to valuable information in this field and laying a solid foundation for future research endeavors.


INTRODUCTION
Artificial intelligence's widespread application in design education has become a developmental trend [1].Design education needs to transform to meet the new demands of the 21st century [2] [3].Donald Schön, a contemporary American educator and a prominent advocate of reflective teaching, proposed that individuals continuously reflect and reframe meaning in the changing new environment until they adopt new frameworks through frame experiments [4].Donald Arthur Norman, an American cognitive psychologist, explains why design education must change by highlighting three aspects, the lack of systematic understanding of behavioral sciences among educators, the inadequacy of curriculum offerings in schools, and the absence of high-quality design practice environments.It is noteworthy that he emphasizes the necessity for change in design education.Norman suggests that designers' skills do not align well with modern society because of the lack of requirements for science, mathematics, technology, and social sciences courses in design education.Furthermore, Norman points out the direction for the change in design education, which is the need for the next generation of designers to understand science and technology, the ability to work across disciplines, and knowledge of human beings, business, and technology, as well as appropriate methods to validate requirements [5].A typical model of design education curriculum should incorporate the structure of art, science, and technology [6].
The first Artificial Intelligence (AI) conference was held at Dartmouth College in 1956, introducing the concept of artificial intelligence.Since then, AI, as a rapidly developing field, has found widespread application in various sectors such as healthcare, automotive, banking and finance, security, and social media.Integrating artificial intelligence and art design can diversify the thinking processes of artistic design and enable future art design to serve humanity more intelligently and personally [7].Artificial intelligence has already been adopted in the education sector or educational institutions to enhance the quality and efficiency of teaching activities [8].However, the related intelligent systems are primarily designed for businesses and individuals rather than specifically for education [9].The theoretical research on integrating the Artificial Intelligencebased Creative Thinking Skills Analysis Model (AI-CTSAM) in the art education curriculum, exploring more possibilities, is ongoing [10].Although numerous studies have examined the combination of artificial intelligence and education, research on integrating artificial intelligence and design education is still nascent.This study builds upon prior research to delve deeper into this area and offer valuable qualitative insights.Within this context, the study aims to address the following two questions: RQ1: What is the current state of literature research and research hotspots?

VISUALIZATION RESULTS ANALYSIS
We sourced the literature for this study from the Science Citation Index Expanded (SCI-EXPANDED), Social Sciences Citation Index (SSCI), and Arts & Humanities Citation Index (AHCI) within the Web of Science Core Collection database.We conducted the search using the following criteria: Topics = ("Design Education" or "Art Education") and (Artificial Intelligence).We limited the retrieval time from January 2013 to May 2023.After excluding irrelevant literature, we obtained a total of 810 relevant articles.We utilized CiteSpace version 6.In this study, the clustered charts yielded a Q-value of 0.4861 and an S-value of 0.7599, indicating a significant and convincing clustering structure.

Country network analysis
Firstly, a national network analysis was conducted(Figure 1).The colors of the citation rings in the national network indicate that the lighter the color, the more recent the corresponding citations.Most of the literature belongs to a relatively recent period.Moreover, from 2013 to 2023, there is an increasing trend of attention toward theoretical research in the field from countries such as Mainland China, the United States, the United Kingdom, Spain, Taiwan (China), Australia, Germany, and South Korea.

Institutional research field analysis
Firstly, we conducted an index of institutions mentioned in the literature.The institution with the highest citation count was RLUK-Research Libraries UK (2018), with 15 citations.The second was Harvard University (2017), with 14 citations.The third was the University of London (2019), with 13 citations.The fourth was the Chinese University of Hong Kong (2018), with 13 citations.The fifth Furthermore, we clustered the results of the institutional indexing according to subject categories and obtained the clustering results.(Figure 2).The largest cluster, labeled as (#0), includes computer science, theory, and methods.The second largest cluster, labeled as (#1), encompasses education and education research.The remaining clusters cover various disciplines, including psychology, engineering, law, and image science.The visual representation demonstrates that integrating artificial intelligence and design education has expanded beyond exploring technical aspects and now showcases a more diverse range of research perspectives.

Burst keyword analysis
The visualization in the figure shows the keyword burst analysis of the top 20 most cited keywords (Figure 3)."Computer-aided instruction" and "computer programming" were introduced in 2013.The topic of children has received sustained attention from 2014 to 2020, with an intensity of 2.16."AI education" started to gain widespread attention in 2020."Health" is the keyword with the highest intensity, reaching 3.04."Motivation" emerged as a research hotspot after 2021.Additionally, keywords such as "model, " "attention, " "comprehension, " "fuzzy logic," "artificial neural networks, " "design, " "classification, " "context, " "behavior, " "outcome, " "people, " "adoption, " all experienced bursts between 2013 and 2023, representing directions that scholars widely focused on during specific periods.

Cluster identification analysis
In order to understand representative literature clusters, we constructed a co-citation network comprising 810 articles from 2013 to 2023-CiteSpace clusters closely related keywords together, which facilitates scholars in analyzing and studying these keywords.The network in this study consists of 10 clusters, numbered from 0 to 9. The smaller the number of these 10 clusters, the more keywords are included in the cluster.The most significant nine clusters are as follows (Figure 4), Cluster labels with smaller ordinal numbers  indicate more members inside the cluster, and at the same time, the color blocks will be larger and relatively darker.It should be emphasized that this study only conducts a specific analysis of the 0-8 clusters with the greatest influence in the cross-documentation literature on artificial intelligence (AI) and design education.
The largest cross-clustering of artificial intelligence and design education (#0) has 57 members and a silhouette value of 0.774.It is labeled as center positioning strategy by LLR, artificial intelligence by the most cited members of this cluster are artificial intelligence (337 times), education (89 times), and post design (80 times).The principal cited article in this cluster is (Wu et al., 2015) [11], which highlights the AI capability and analyzes how AI FDSS can intervene to develop a comprehensive center location strategy and its application in Dr I-Kids Education Center.The data collection of this cluster can help stakeholders who want to find applications to expand AI in the field of design education to quickly find the strategy models and reference methods proposed in the current literature.
The second largest cross-cluster of artificial intelligence and design education (#1) has 54 members and a silhouette value of 0.778.It is labeled as machine learning by LLR, artificial intelligence by LSI, and playwriting technique (1.49) by MI.It is labeled as machine learning by LLR, artificial intelligence by LSI, and playwriting technique (1.49) by MI.The most cited members of this cluster are machine learning (62 times), deep learning (42 times), and big data (39 times).The top-cited article in this cluster is (Bourechak et al., 2023) [12].This literature analyzes the convergence of artificial intelligence and edge computing in eight domains: Smart Agriculture, Smart Environment, Smart Grid, Smart Healthcare, Smart Industry, Smart Education, Smart Transportation, and Security and Privacy, from a global perspective.
The third largest cross-cluster of artificial intelligence and design education (#2) has 41 members and a silhouette value of 0.679.It is labeled as a college student sport by LLR and artificial intelligence by LSI.The most cited members of this cluster are knowledge (21 times), artificial intelligence (ai) (16 times), and neural networks (11 times).The principal cited article in this cluster is (Yoon & Kim, 2015) [13], which verifies the positive impact of game-based learning on teaching methods, based on the Angry Birds AI competition.
The fourth largest cross-cluster of artificial intelligence and design education (#3) has 36 members and a silhouette value of 0.684.It is labeled as ai-based learning style prediction by LLR, high school student grade by LSI, and cad platform (0.17) by MI.The most cited members of this cluster are learning analytics (14 times), data mining (12 times), and teachers (10 times).The principal cited article in this cluster is (Raffaghelli et al., 2022) [14], which emphasizes the use of artificial intelligence algorithms by early warning systems to predict learner behavior, and the results of the article suggest that the introduction of artificial intelligence in the familiarization phase seems to be essential for student acceptance in virtual classrooms in higher education.
The fifth cross-cluster of artificial intelligence and design education (#4) has 35 members and a silhouette value of 0.752.It is labeled as cognitive load by LLR, artificial intelligence by LSI, and students' intention (0.41) by MI.The most cited members of this cluster are students (35), motivation (13), and decision making (10).The principal cited article in this cluster is (Lin et al., 2021) [15], which analyzes the AI learning of observers by perceived usefulness, perceived ease of use, subjective norms, AI attitudes, and behavioral intentions.
The sixth cross-cluster of artificial intelligence and design education (#5) has 34 members and a silhouette value of 0.775.It is labeled as educational robotics by LLR and artificial intelligence by LSI.The most cited members of this cluster are technology (38 times), virtual reality (19 times), and augmented reality (16 times).The principal cited article in this cluster is (I.Gaudiello, 2012) [16], which analyzes the state and specific learning styles according to different AI robots to achieve educational goals and presents the challenges to integrating AI into school curricula.
The seventh cross-cluster of artificial intelligence and design education (#6) has 34 members and a silhouette value of 0.652.It is labeled as enabling technologies challenge by LLR, artificial as enabling technologies challenge by LLR, artificial intelligence by LSI, and artificial intelligence capability (0.3) by MI.The most cited members of this cluster are management (23 times), internet (16 times), and systems (13 times).The principal cited article in this cluster is (Fuller et al., 2020) [17], which identifies the challenges within the field of artificial intelligence and deep learning, and classifies machine learning into supervised learning, unsupervised learning, and deep learning.
The eighth cross-cluster of artificial intelligence and design education (#7) has 32 members and a silhouette value of 0.935.It is labeled as using sensing technologies by LLR, empowering the classroom.It is labeled as using sensing technologies by LLR, empowering classroom observation by LSI, and artificial intelligence (0.07) by MI.The most cited members of this cluster are children (10 times), information (9 times), and natural language processing (8 times).The principal cited article in this cluster is (Huang et al., 2014) [18], which aims to enhance the effectiveness of teachers' classroom observation through the study of sensing technology in an e-book reading behavior monitoring system, and the results of the study show the usability and functionality of the proposed system to be effective.
This study presents the smallest, ninth cross-clustering of artificial intelligence and design education (#8) has 28 members and a silhouette value of 0.716.It is labeled as a multidisciplinary perspective by LLR.Educational robots by the most cited members of this cluster are higher education (32 times), impact (30 times), and future (15 times).The principal cited article in the cluster is (Dwivedi et al., 2023) [19], which highlights some of the most transformative impacts that teaching, learning, and scholarship will experience in the era of artificial intelligence by exploring ChatGPT, a generative conversational artificial intelligence.As well as proposing insights such as the critical importance of identifying and implementing policies to prevent the misuse and abuse of generative AI.

DEVELOPMENT TRENDS
The combination and intersection of artificial intelligence(AI) and design education is an interdisciplinary field of innovative research.The visual map formed during the timeframe of this study (Figure 5) reveals that from 2013 to 2017, research focused on significant clusters, particularly Cluster #0, Cluster #1, and Cluster #5.These clusters revolved around the center positioning strategy, machine learning, and educational robotics, indicating the exploration and acquisition of expertise, technologies, machines, skills, and models in this field.From 2018 onwards, research directions gradually diversified, leading to a more even distribution of research interests across nine core clusters.The center positioning strategy has matured, resulting in a decreasing trend in Cluster #0.The diverse research focuses indicate a growing interest among scholars in applying artificial intelligence in design education, facilitating rapid theoretical and methodological advancements, and laying the groundwork for practical implementations.
The future integration of artificial intelligence and design education will involve multiple perspectives from teachers, students, schools, and stakeholders invested in artificial intelligence [20].The emerging trends indicate that with the support of artificial intelligence, various aspects of design education, including hardware infrastructure, software technologies, instructional models, practical applications, and feedback assessment, will develop in an integrated manner [21].

CONCLUSION
Through the analysis of 810 highly relevant literature on artificial intelligence and design education in the Web of Science Core Collection database, this study draws the following conclusions: Firstly, the study examines the literature's current status and research trends.We can observe that the corresponding citations are recent by analyzing the colors of the national network citation rings.There is a high level of interest in integrating artificial intelligence and design education in countries such as China, the United States, the United Kingdom, Spain, Australia, Germany, and South Korea.Moreover, the theoretical research in this field is showing an increasing trend year by year.The institutional network analysis reveals that Research Libraries UK, Harvard University, University of London, Chinese University of Hong Kong, University of California System, N8 Research Partnership, University System of Ohio, Beijing Normal University, Stanford University, and Nanyang Technological University & National Institute of Education (NIE) Singapore have demonstrated significant interest in this field.Through the analysis of high-frequency keywords and keyword burst visualization, the study identifies 20 key terms that are relatively important in the field of combining artificial intelligence and design education, including computer-assisted instruction, computer programming, children, model, attention, comprehension, fuzzy logic, artificial intelligence (AI), artificial neural networks, autonomous robots, and design.These keywords represent the research hotspots in this field.
Secondly, the study addresses the future trends of combining artificial intelligence and design education.It identifies nine significant clusters: center positioning strategy, machine learning, college student sport, AI-based learning style prediction, cognitive load, educational robotics, enabling technologies challenge, using sensing technologies, and a multidisciplinary perspective.These clusters provide a theoretical basis for further research in developing evaluation systems in this field.The temporal segmentation in Figure 5 further reveals the research trends in the field and how the integration of artificial intelligence and design education has evolved.
Finally, this study's unique value lies in the significant literature obtained through the natural clustering of software without manual intervention, which enhances its reference value.Additionally, the visual analysis conducted in this research allows researchers in related fields to quickly and effectively obtain valuable information by combining artificial intelligence and design education, providing a solid foundation for subsequent studies.
2.2 to analyze the literature in this study.This version is highly stable, ensuring adequate support for the literature analysis conducted in this research.Regarding the analysis of the results from the visualization charts in CiteSpace, two values are essential for the clustered outcomes: Q-value and S-value, which indicate the quality of clustering.Modularity (Q-value) represents the module value of clustering, with Q>0.3 generally considered a significant clustering structure.Silhouette (S-value) represents the average silhouette value of clustering, with S>0.5 indicating reasonable clustering and S>0.7 indicating convincing clustering.

Figure 2 :
Figure 2: Subject Category of Institution Visualized.

Figure 3 :
Figure 3: Top 20 Keywords with the Strongest Citation Bursts.