Integrating Creative Problem Solving and Generative AI in Animation Education: Advancing Sustainability-Related Competencies in Higher Education

Abstract

This study examines how integrating Creative Problem Solving (CPS) and generative artificial intelligence (GenAI) within animation storytelling education can foster sustainability-related competencies in higher education. A twelve-week mixed-methods action research design was implemented in a “Storytelling and Scriptwriting” course at a university of technology in northern Taiwan (N = 60). The intervention design combined a CPS-aligned instructional sequence, six scaffolded assignments (including a text-to-image resemiotization task), pre–post CPS cognition and affect scales, CPS-dimensioned assignment self-assessments, reflective journals, and expert evaluations of final story prototypes using the Consensual Assessment Technique. Quantitative results showed significant gains in students’ CPS-related narrative cognition and affective resilience (p < 0.001), as well as consistently high self-reported engagement across CPS dimensions for all assignments, particularly for the text-to-image and personal narrative tasks. Expert ratings indicated high levels of originality, narrative coherence, emotional impact, and social relevance in final prototypes, while qualitative data highlighted reduced “blank page” anxiety, greater willingness to revise, and more collaborative, systems-oriented narrative reasoning. The findings suggest that a CPS- and GenAI-supported teaching model can function as a cognitive bridge for heterogeneous cohorts, positioning GenAI as a conditional amplifier embedded within a reflective CPS framework and helping to translate abstract sustainability-related competencies—such as anticipatory, normative, strategic, and interpersonal competencies—into concrete creative media practices.

1. Introduction

1.1. Background and Motivation

In globalized and digitalized economies, national competitiveness increasingly depends on a workforce capable of creative and higher-order thinking rather than technical proficiency alone. Creativity has been widely recognized as a foundational component of innovation and as a critical internal resource enabling individuals to address complex, ill-structured problems [1,2]. In particular, research on creative self-beliefs suggests that students’ confidence in their creative abilities plays a pivotal role in whether they engage with complex, open-ended problems in the first place [1]. Studies on creative science teaching in East Asian contexts similarly indicate that limited pedagogical support for creativity and problem finding constrains students’ ability to move beyond procedural task completion [3]. Consequently, higher education institutions are under growing pressure to move beyond knowledge transmission and toward learning environments that intentionally cultivate creativity, critical thinking, and problem-solving capacities.

International policy frameworks reflect this paradigm shift. The Organization for Economic Co-operation and Development (OECD) identifies creativity and critical thinking as core competencies for twenty-first-century education and provides pedagogical guidelines for embedding creativity-oriented tasks within authentic classroom contexts [2]. Within Education for Sustainable Development (ESD), sustainability-related competencies have been conceptualized as interrelated capacities that enable learners to address complex socio-ecological challenges in a reflective and action-oriented manner. Building on an extensive review of higher education programs, Wiek et al. [4] proposed a widely cited framework comprising systems-thinking, anticipatory, normative, strategic, and interpersonal competencies as key learning goals for sustainability-oriented curricula. Subsequent work has further elaborated and updated this framework, identifying emerging intrapersonal and implementation competencies and synthesizing how different pedagogical approaches seek to operationalize these abstract competencies in practice [5,6]. These competencies enable learners to navigate uncertainty, engage with societal complexity, and contribute to sustainable transformation.

Such developments are particularly significant for disciplines traditionally centered on technical skill acquisition, including animation and digital media production. While technical expertise remains essential, sustainable development in creative industries increasingly demands originality, narrative sophistication, and cultural responsiveness.

In Taiwan, the animation industry historically functioned as an outsourcing base within global production networks, accumulating extensive technical capabilities but comparatively limited original intellectual property (IP) development [7]. As the industry transitions toward IP-driven content creation, professional expectations have shifted toward higher-order narrative competencies, including story conceptualization, character development, and thematic coherence. However, higher education programs in design and media frequently prioritize software proficiency and production techniques, offering limited structured support for narrative creativity and problem-finding skills.

From a cognitive psychology perspective, difficulties in generating original story concepts often reflect underdeveloped problem-finding skills—an essential component of creative performance [8,9]. This challenge is particularly salient among Generation Z learners, who are immersed in digital audiovisual environments but often lack explicit frameworks for constructing narrative structures or aligning storytelling with contemporary social issues [3,10]. At the same time, recent industry and media analyses suggest that Gen Z audiences increasingly value animation and digital content that foreground authentic friendship dynamics and relatable everyday experiences rather than purely spectacular visuals [11,12]. Addressing this gap requires pedagogical models that bridge technical animation skills with higher-order narrative reasoning while explicitly supporting selected sustainability-related competencies such as reflective judgment, collaboration, and creative problem solving [13,14].

Creativity research has long emphasized the importance of problem finding as a precursor to innovative outcomes [8]. Within Amabile’s [15] componential model, creativity emerges from the interaction of domain-relevant skills, creativity-relevant processes, and intrinsic motivation. In animation education, technical proficiency may satisfy domain-relevant skills, yet structured creativity-relevant processes—such as Creative Problem Solving (CPS)—are often underdeveloped. This imbalance may partially explain why technically competent students struggle with original narrative generation.

Integrating sustainability-related competencies into creative media education may provide a novel pathway for aligning artistic practice with sustainability-oriented higher education. By situating narrative problem finding, character development, and multimodal storytelling within a CPS framework informed by ESD competencies, educators may support both creative performance and sustainability-related learning outcomes [4,5].

1.2. Research Gap and Rationale

The Creative Problem Solving (CPS) model offers a structured yet flexible framework for addressing complex challenges. Originating in Osborn’s early work, CPS emphasizes iterative cycles of divergent and convergent thinking across phases such as problem clarification, idea generation, solution development, and implementation planning [16,17]. Contemporary adaptations conceptualize CPS as a dynamic and domain-adaptable model that provides cognitive scaffolding for ill-structured problem contexts [18].

Empirical research demonstrates that CPS-based instruction enhances university students’ creative thinking and academic performance across disciplines, including statistics, mathematics, and STEM education [19]. However, its application within animation storytelling and scriptwriting education remains limited. Existing CPS research primarily focuses on well-defined analytical tasks, whereas animation storytelling involves multimodal, affective, and narrative integration under conditions of ambiguity [9].

Traditional animation pedagogy frequently relies on established structural templates such as the hero’s journey or folktale morphology [20,21]. Although these frameworks provide useful narrative scaffolds, they may constrain creative exploration in contemporary media environments characterized by nonlinear storytelling, interactive systems, and transmedia narratives [22]. Furthermore, few studies provide empirically validated teaching models that systematically integrate CPS processes with animation-specific narrative design.

The rapid emergence of generative artificial intelligence introduces both opportunities and challenges in creative education [23]. AI tools can function as cognitive collaborators, supporting rapid ideation, visualization, and iterative refinement [24,25,26]. At the same time, they raise critical concerns regarding authorship, ethical practice, and overreliance on automated outputs [23,27]. Current research on AI in education predominantly examines text-based tasks or general design ideation, with limited empirical investigation into how CPS and AI can be jointly orchestrated in animation storytelling contexts to foster sustainability-related competencies such as metacognitive regulation, strategic thinking, and collaborative meaning making.

Three research gaps can therefore be identified. First, empirically grounded CPS- and AI-supported teaching models tailored specifically to animation storytelling in higher education remain scarce. Second, visual thinking and multimodal transformation processes are rarely operationalized into structured, stepwise assignments that support novice animators in translating textual concepts into coherent visual narratives [28,29,30]. Third, the role of AI tools as structured supports within CPS phases has not been systematically examined from a sustainability education perspective, particularly regarding their potential to function as conditional amplifiers of creativity while preserving students’ reflective agency [4,5,31].

This study addresses these gaps by developing and evaluating a CPS- and AI-supported teaching model within an animation storytelling course, investigating its effects on narrative cognition, creative problem solving, and selected sustainability-related competency development.

1.3. Research Objectives and Questions

The core research problem addressed here is the perceived gap between technical efficiency and narrative depth in AI-supported creative education. This study investigates whether a CPS framework can bridge this gap by providing metacognitive scaffolding that helps students to navigate the generative power of AI without sacrificing critical storytelling logic.

This study adopts an action research approach in a storytelling and scriptwriting course offered within a digital multimedia design program at a university of technology in northern Taiwan. Sixty undergraduates from diverse disciplinary backgrounds participated in a twelve-week intervention integrating a dual-track CPS- and AI-supported teaching model with scaffolded assignments and AI-supported tools.

The study pursues the following objectives:

To construct a CPS-based, structurally explicit teaching model for animation storytelling that can be implemented in higher education contexts and support selected sustainability-related competencies.

To examine the effects of this model on students’ narrative cognition, affective resilience in creative problem solving, and perceived cognitive load.

To investigate how CPS-informed assignments and AI-supported tools influence students’ ability to transform personal experiences into socially meaningful narrative prototypes aligned with sustainability-related competencies.

Accordingly, the study addresses three research questions:

RQ1: How does integrating a CPS- and AI-supported teaching model influence students’ narrative cognition and affective orientations toward creative problem solving?

RQ2: How do CPS-structured assignments affect students’ perceived efficiency in transforming textual concepts into visual story prototypes and their subjective cognitive load?

RQ3: To what extent is the CPS- and AI-supported teaching model associated with improvements in students’ creative thinking and reductions in perceived creative blockages?

By answering these questions, this study contributes an empirically grounded pedagogical framework that integrates creative media education with sustainability-oriented competency development, thereby extending the scope of Education for Sustainable Development into animation and digital storytelling practices.

2. Literature Review

2.1. The Evolution of Creative Problem Solving and Solution Development

The Creative Problem Solving (CPS) model was originally proposed by Osborn, who conceptualized creativity as a structured process alternating between divergent and convergent thinking to generate and refine ideas [16]. Subsequent refinements by Isaksen and colleagues developed CPS into a multi-stage framework—including problem clarification, idea generation, solution development, and implementation—widely applied in organizational innovation and educational contexts [17,32].

Early formulations were criticized for linearity and limited responsiveness to real-world complexity. Later reconceptualizations reframed CPS as a cyclical and iterative process, allowing stages to be revisited as new insights emerge and problem definitions evolve [17]. This evolution aligns closely with contemporary sustainability challenges, which are characterized by uncertainty, systemic interdependence, and non-linear dynamics [4,5].

Empirical studies demonstrate that CPS-based instruction enhances creative thinking and problem-solving abilities across disciplines. Research in statistics and mathematics education reports significant improvements in learning outcomes and creative performance following CPS interventions [18,19]. Similar findings have been observed in STEM and teacher education contexts, where CPS supports higher-order reasoning and innovation-related competencies [19].

Within animation education, the solution development phase of CPS is particularly critical. As digital media evolve toward interactive, nonlinear, and multi-path narrative systems, animators increasingly design narrative possibility spaces rather than single, fixed storylines [22]. Traditional frameworks such as the hero’s journey and folktale morphology provide stable structural templates but may constrain creative exploration when students are required to construct branching plots or interactive systems [20,21].

Despite CPS’s empirical support, few studies systematically map CPS phases onto animation pre-production processes such as character construction, plot architecture, and narrative path engineering. Given that sustainability-oriented education emphasizes systems thinking and anticipatory competence [4,5], adapting CPS to animation storytelling offers a promising approach to integrating creative media education with sustainability competency development. In this study, CPS is therefore treated not merely as a generic creativity heuristic, but as a structured cognitive framework that can be mapped onto specific narrative development tasks in animation storytelling.

Within Education for Sustainable Development (ESD), sustainability-related competencies have been conceptualized as interrelated capacities that enable learners to address complex socio-ecological challenges in a reflective and action-oriented manner. Building on an extensive review of higher education programs, Wiek et al. [4] proposed a widely cited framework comprising systems-thinking, anticipatory, normative, strategic, and interpersonal competencies as key learning goals for sustainability-oriented curricula. Subsequent work has further elaborated and updated this framework, identifying emerging intrapersonal and implementation competencies and synthesizing how different pedagogical approaches—such as project-based learning, problem-based learning, and community-engaged courses—seek to operationalize these abstract competencies in practice [4,5,6].

Despite this growing body of scholarship, relatively few studies have examined how creative media practices, and particularly animation storytelling, can be systematically aligned with sustainability-related competencies through structured instructional models. This study responds to that gap by integrating CPS and GenAI-supported visual resemiotization into an animation storytelling course, with the explicit aim of cultivating sustainability-related competencies such as anticipatory thinking, normative reflection, and collaborative problem solving.

2.2. Visual Thinking and Resemiotization

Arnheim [28] argued that visual perception constitutes structured thinking rather than passive reception, positioning visualization as a cognitive process that externalizes and tests abstract concepts. Similarly, Hayes and Flower conceptualized writing as a recursive problem-solving activity involving planning, translating, and revising, supported by external representations that interact with internal cognition [33].

McKim [29] proposed a three-stage cycle of seeing–imagining–drawing, highlighting the dynamic interplay between observation, internal imagery, and graphic production. In animation education, this cycle underpins core practices such as character sketching, storyboard development, and compositional studies [30].

From a social semiotic perspective, the transformation of verbal concepts into multimodal forms can be understood as resemiotization—a process through which meaning shifts across semiotic modes. Multimedia learning research further demonstrates that well-designed visual supports can reduce cognitive load and enhance deep processing and creative engagement [34]. Multimodal discourse studies in digital humanities similarly illustrate how text–image transformations can support meaning-making and engagement in AI-mediated cultural heritage projects [25,35].

However, much of this literature focuses on general multimedia learning or multimodal communication rather than structured integration into animation scriptwriting workflows. Few empirical studies operationalize visual thinking into scaffolded assignments that explicitly support novice animators in translating textual prompts into coherent visual narrative systems [30].

To address this gap, the present study integrates CPS idea generation and solution development with structured visual transformation tasks. A text-to-image assignment operationalizes resemiotization within CPS phases, supporting learners in externalizing abstract ideas while managing cognitive load. This approach aligns with sustainability-oriented pedagogy by enhancing reflective thinking and metacognitive awareness in creative production [5]. In the present study, resemiotization is operationalized through a text-to-image assignment that guides students from verbal prompts to visual narrative prototypes while encouraging iterative reflection.

2.3. Character Arcs as Cognitive Reframing Processes

Character transformation constitutes a central feature of compelling animation narratives. McKee [36] argues that effective storytelling revolves around shifts in characters’ values, motivations, and self-concepts under pressure. Interpreted through a CPS lens, a character’s initial predicament parallels problem definition; subsequent struggles reflect idea generation and evaluation; and eventual transformation corresponds to solution resolution.

Redvall [9] emphasizes that scriptwriting involves iterative, constraint-based decision-making shaped by feasibility, resources, and audience expectations. Screenwriters frequently employ what-if questioning strategies, resonating with divergent thinking principles within CPS [32].

Contemporary media environments increasingly prioritize authenticity, identity exploration, and socially grounded storytelling over formulaic heroism. Audience research indicates that identification with characters—particularly those reflecting viewers’ lived experiences—enhances engagement and emotional investment [37,38,39]. Recent industry and media analyses suggest that Gen Z audiences increasingly value animation and digital content that foreground authentic friendship dynamics and relatable everyday experiences [11,12].

Despite extensive narrative theory research, relatively few studies conceptualize character arcs as structured cognitive frameworks that can be explicitly taught using problem-solving models. Aligning CPS stages with character arc development may strengthen students’ ability to frame conflicts, evaluate narrative alternatives, and construct meaningful growth trajectories [9].

From a sustainability education perspective, character arcs can also serve as metaphorical representations of cognitive reframing and transformative learning, fostering normative competence and reflective judgment [4,5]. Embedding CPS within character construction thus links narrative pedagogy with sustainability-related competency development.

2.4. Integrating CPS and GenAI as a Scaffold for Animation and Sustainability Education

The integration of generative artificial intelligence (GenAI) into creative education requires careful theoretical positioning. Recent work in design, architecture, and advertising education shows that GenAI tools can expand students’ ideation space and support rapid iteration, yet simultaneously raise concerns about originality and ethical practice [23,24,26,27]. From the perspective of componential theories of creativity [15], GenAI influences domain-relevant skills and task motivation by providing instant exemplars and alternative framings. However, systematic evidence suggests that GenAI functions as a conditional amplifier [31,40]: while it can enhance performance for novices, it also risks reducing the collective diversity of outputs unless counterbalanced by pedagogical designs that preserve critical judgment and metacognitive regulation.

In this study, the theoretical justification for linking CPS to animation storytelling lies in the structural alignment between CPS phases and the cognitive demands of the narrative development process [17]. Animation scriptwriting is fundamentally an ill-structured problem-solving task, moving from an abstract concept to a structured sequence. We argue that the interaction between CPS and GenAI directly influences narrative cognition, affective resilience, and creative performance—through a process of metacognitive scaffolding [31]. Specifically, this study conceptualizes GenAI not as a surrogate creator, but as a scaffolded cognitive perturbation device [31] situated within the CPS framework.

The Clarify phase targets narrative coherence by encouraging students to define story problems and thematic goals. The Ideate phase utilizes GenAI to expand the visual possibility space and support resemiotization. Subsequently, the Develop and Implement phases function as quality-control mechanisms, ensuring that AI-generated outputs align with narrative logic and sustainability-oriented judgment [4,5,6]. This logical chain ensures that students do not passively adopt AI suggestions but instead use them as a catalyst for deeper systemic and ethical reflection. By navigating the generative power of AI through this structured CPS framework, the model provides a foundation for fostering the higher-level competencies required in contemporary animation education.

The theoretical alignment between these CPS phases, the functional roles of GenAI, and their corresponding pedagogical objectives in animation education is summarized in

Table 1. Conceptual Mapping of CPS Phases, GenAI Roles, and Animation Learning Outcomes.

CPS Phase	Core CPS Function	Role of GenAI	Animation Learning Outcomes
Clarify	Defining narrative “pain points” and story goals.	Analytical Support: Assisting in research and context analysis.	Enhancing narrative coherence and thematic depth.
Ideate	Generating diverse narrative possibilities.	Divergent Tool: Generating character/setting prototypes (Resemiotization) [29].	Expanding creative breadth and visual imagination.
Develop	Selecting and optimizing AI-generated outputs.	Iterative Tool: Rapidly refining and modifying visual drafts.	Strengthening critical thinking and narrative detail.
Implement	Final visual transformation and execution.	Production Tool: Generating final keyframes and multimodal assets.	Improving artifact quality and technical integration.

.

As summarized in Table 1, this alignment positions GenAI as a scaffolded support within the CPS process, with each phase associated with a distinct instructional function and learning focus. Rather than serving as an autonomous creator, GenAI is thus integrated into a structured sequence of narrative development tasks that supports students’ creative reasoning and connects classroom practice to the sustainability-related competencies targeted in this study.

2.5. Research Hypotheses

The reviewed literature indicates that CPS enhances creative problem solving across disciplines, yet its structured application to animation storytelling remains underexplored [18,19]. Research on visual thinking and resemiotization has seldom been operationalized into scaffolded animation assignments, and AI-supported creativity research rarely integrates CPS within sustainability-oriented educational frameworks [5,25,34].

To address these gaps, this study implements a CPS- and AI-supported teaching model in an animation storytelling course and empirically examines its impact on narrative cognition, transformation efficiency, and creative performance.

Accordingly, the following hypotheses are proposed:

H1. 

Students’ CPS-related narrative cognition scores will significantly increase following participation in the CPS- and AI-supported teaching model.

H2. 

Integrating structured visual resemiotization strategies within CPS assignments in the CPS- and AI-supported teaching model will reduce students’ perceived cognitive load and enhance perceived efficiency in transforming textual concepts into visual story prototypes.

H3. 

Students’ final story prototypes will demonstrate significantly higher creative quality, as assessed through expert consensual ratings, following the CPS- and AI-supported teaching model.

3. Methods

3.1. Research Design

This study adopted a mixed-methods action research design to examine the effects of a CPS- and AI-supported teaching model in a higher education animation storytelling course. The action research perspective allowed the instructor–researcher to iteratively refine the innovative pedagogical intervention within an authentic, complex, and heterogeneous classroom context [41,42]. The design combined quantitative pre- and post-tests, CPS-dimensioned assignment self-assessments, and qualitative reflective journals, supplemented by independent expert evaluations of students’ final story prototypes using the Consensual Assessment Technique (CAT).

Given the formal credit-bearing nature of the course and the commitment to pedagogical equity [43], a non-intervention control group was not established so that all students had equal access to the instructional resources and the AI-supported learning environment. The CPS- and AI-supported intervention was implemented over a twelve-week semester and was structured around two intertwined tracks: (1) in-class instruction aligned with CPS phases and core concepts of animation storytelling, and (2) a sequence of six scaffolded assignments designed to operationalize CPS and visual resemiotization strategies in students’ creative processes. This dual-track design was intended to support both conceptual understanding and practical application of CPS in narrative development. By collecting data systematically across these six stages, the study aimed to capture the incremental developmental trajectory of students’ creative cognition, providing a process-oriented validation of the intervention’s effectiveness.

3.2. Participants and Context

The participants were 60 undergraduate students enrolled in a course titled “Storytelling and Scriptwriting” at a university of technology in northern Taiwan. The cohort was characterized by highly heterogeneous academic backgrounds: 36 students were majoring in digital multimedia design, 8 in visual communication design, 6 in business, 2 in film, and 8 from other departments. Most students were in their first year of study.

Notably, over 85% of the participants reported having no prior formal training in narrative scriptwriting, story structure, or structured creative problem-solving. While some participants had prior exposure to animation software through vocational high school training or personal interest, this technical familiarity did not extend to narrative theory or systemic creative strategies. This “zero-baseline” in narrative design is crucial for the study’s internal validity; it suggests that the high-level creative outputs later validated by independent experts were unlikely to have resulted from prior academic knowledge, natural maturation, or unguided external AI usage alone. Instead, the observed improvements can be more directly attributed to the specific metacognitive scaffolds [3,33] provided during the 12-week CPS-integrated intervention. All participants provided informed consent, and all data were anonymized to comply with institutional research ethics guidelines.

3.3. CPS-Based Course Design and Implementation

The course design explicitly mapped the cognitive stages of CPS onto core units of animation storytelling through a twelve-week intervention. This pedagogical structure was organized into six instructional units, each lasting two weeks and aligned with a specific CPS phase, as summarized in

Table 2. Instructional units and CPS phase mapping.

Teaching Unit	Unit Title	Key Examples and Strategies	Duration	CPS Phase Mapping
Unit 1	Character Conflict and Construction	Personality mismatch in “Shark Tale”	2 weeks	Clarifying
Unit 2	Suspense Design	“Barber shop” narrative drive case	2 weeks	Problem–Finding
Unit 3	Foreshadowing and Setup	Prop–driven causality in “Up”	2 weeks	Generate Ideas
Unit 4	Theme and Message	Environmental themes in “Nausicaä of the Valley of the Wind”	2 weeks	Idea Finding
Unit 5	Complications and Resolutions	Short–film reversal matrix design	2 weeks	Develop Solutions
Unit 6	Character Growth Curve	Transformation indicators in “Thor”	2 weeks	Implement & Evaluate

. While Table 2 outlines the thematic progression and exemplar materials used (e.g., analyzing narrative drive in Barber Shop or character arcs in Thor), the underlying operational mechanisms that bridge CPS theory and AI-supported practice are further detailed in

Table 3. Operationalization of the CPS-AI Instructional Framework.

CPS Phase	Instructional Mechanism (Teaching Activities)	GenAI Practical Application	Narrative Artifact (Output)
Clarify	“5W1H” mapping and stakeholder analysis of socio-environmental issues (SDGs) [16,17].	Using LLMs (e.g., ChatGPT-5.3) for stakeholder empathy mapping and theme exploration.	Problem Statement & Theme.
Ideate	“SCAMPER” brainstorming for character arcs [16,17].	Text-to-Image (e.g., Leonardo.ai) for visual prototyping and character resemiotization.	Character Bios & Concept Art.
Develop	Peer-review via “Plus-Minus-Interesting” (PMI) framework for iterative refinement [16,17].	Image-to-Video for testing scene transitions.	Storyboard & Script Draft.
Implement	Final “Resemiotization” workshop and expert-led CAT evaluation [28,29,30].	Using AI tool chains for final rendering/editing.	Final Animated Prototype.

.

To move beyond a simplistic mapping, the instructional framework utilized GenAI tools not as automated content generators, but as cognitive perturbation devices integrated into specific classroom mechanisms. As shown in Table 3, each CPS phase was operationalized through targeted teaching activities and corresponding AI applications.

For example, in the “Clarify” phase, students employed “5W1H” mapping to identify socio-environmental issues (SDGs), using Large Language Models (LLMs) to perform stakeholder empathy mapping. During the “Ideate” phase, the SCAMPER brainstorming technique was paired with text-to-image tools to externalize character prototypes. Crucially, the “Develop” phase incorporated peer-review feedback using the “Plus-Minus-Interesting” (PMI) technique, where AI image-to-video tools helped students visualize and test scene transitions before finalizing their scripts. This layered approach ensures that GenAI supports, rather than replaces, the rigorous analytical and creative demands of the CPS process [23,24,26].

To provide a granular roadmap of the pedagogical intervention, the 12-week instructional sequence was structured around the four primary phases of the CPS framework, integrated with specific GenAI-supported tasks (see Table 3).

The implementation was divided into three core stages: (1) Weeks 1–3 (Preparation & Clarify) focused on identifying sustainability themes; (2) Weeks 4–8 (Ideate & Develop) involved iterative character design and storyboarding using AI toolchains; and (3) Weeks 9–12 (Implement & Evaluate) culminated in a resemiotization workshop and final prototype production.

3.4. Scaffolded Assignments and Cognitive Scaffolding

To translate the CPS framework into students’ concrete creative practice, the course incorporated six stepwise assignments that followed a progressively complex logic, integrating technical narrative skills with specific sustainability-oriented thematic challenges (Table 3), from observation to original prototype development (Figure 1). The sequence was designed to lower initial cognitive load for novices while gradually increasing demands on narrative integration and creative problem solving [34].

The six assignments were:

Character conception (observation and analysis): Students analyzed characters from existing animations, identifying conflicts, goals, and personality traits, thereby practicing fact finding and problem framing.

Suspense creation (event imitation): Students designed suspenseful situations by adapting and recontextualizing narrative patterns from provided cases, developing sensitivity to problem finding and narrative tension.

Text-to-image (guided resemiotization): Students received short textual descriptions and were asked to generate corresponding visual panels or sketches, practicing resemiotization by transforming verbal cues into visual narrative sequences [28,29].

“Tiger Aunt” story rewriting: Students creatively rewrote a classic folk-tale scenario, employing divergent thinking to explore alternative plotlines, character motivations, and endings.

“Three Regrets” personal narrative: Students drew on their own life experiences to construct short stories centered on unresolved regrets, encouraging them to link personal emotion with narrative conflict and to address Generation Z’s preference for authenticity and real-life resonance [12,38].

Final story prototype: To ensure a deep application of sustainability themes, students were required to connect their narratives to contemporary socio-cultural dilemmas. Suggested themes included: (1) exploration of women’s autonomy; (2) reflections on the ethics of artificial intelligence; (3) the integration of virtual and real in urban spaces; and (4) the slash between student and professional identities. Integrating the CPS framework from problem definition to solution evaluation, the project demonstrated the application of narrative cognition to address the challenges of sustainable development in the real world.

Each assignment was accompanied by explicit CPS prompts (e.g., clarifying questions, idea listing, criteria for selecting ideas) and visual thinking tools such as mind maps, story grids, and simple storyboard templates [29,30]. The text-to-image assignment (Assignment 3) was particularly designed to operationalize visual resemiotization, helping students overcome blank page anxiety by providing textual anchors that could be transformed into visual sequences, with optional AI-generated sketches serving as additional stimuli [25,35]. This stepwise scaffold was intended to gradually shift students’ cognitive resources from basic idea generation towards higher-level narrative logic evaluation and multimodal integration.

3.5. Measures and Data Collection

To enhance internal validity and mitigate the absence of a control group, this study utilized methodological triangulation [44,45]. We cross-referenced quantitative gains in pre- and post-tests with qualitative evidence from students’ reflective journals and external validation from independent expert evaluations (CAT). This multi-source evidence chain ensures that observed improvements in creative cognition are not merely artifacts of single-source bias but represent a consistent developmental trajectory [46] prompted by the CPS-AI intervention. The specific instruments and data collection procedures are detailed below.

3.5.1. CPS Cognition and Affect Scale

To evaluate changes in students’ CPS-related learning outcomes, this study employed a CPS cognition and affect scale adapted from established frameworks in creativity education [18,19]. The adaptation underwent a rigorous three-stage validation process to ensure its suitability for the animation storytelling context. First, a Content Validity Index (CVI) was established by a panel of three experts in digital media and educational psychology, who evaluated item relevance and clarity, resulting in a CVI of 0.92. Second, a pilot study was conducted with 15 students from a previous cohort to identify and resolve potential ambiguities in phrasing. Third, the final instrument’s internal consistency was verified using Cronbach’s alpha (α) calculated via SPSS v.26 based on the current sample’s variances. Both subscales achieved coefficients above 0.80, confirming high reliability for this specific pedagogical environment.

The instrument comprised two subscales:

1. The narrative cognition subscale: This included eight items requiring students to evaluate changes in their creative problem solving ability across core story elements, such as character conflict, suspense techniques, thematic messaging, and scene outline writing.

2. The affective attitude subscale: This contained three items targeting confidence in creative problem solving for scriptwriting, persistence in information seeking during complex challenges, and emotional regulation when facing uncertainty or temporary failure.

All items were rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The scale was administered at the beginning (pre-test) and end (post-test) of the twelve-week course. Pre–post differences were analyzed to address RQ1 and test Hypothesis H1 regarding the impact of the CPS- and AI-supported teaching model on students’ CPS-related competencies.

The adaptation followed standard academic protocols for non-commercial research; the original instruments [18,19] are in the public domain for educational use, and their modification was strictly limited to domain-specific terminology (e.g., replacing ‘general problem’ with ‘narrative conflict’) to maintain the theoretical integrity of the constructs while ensuring ecological validity.

3.5.2. CPS Self-Assessment of Assignments

For each of the six core assignments, students completed a brief self-assessment form in which they rated their performance on five CPS dimensions: fact finding, problem finding, idea finding, solution finding, and acceptance finding. Ratings were given on a seven-point Likert scale (1 = very low, 7 = very high). These self-assessments were used to examine how students perceived their CPS engagement across different tasks and to identify which assignments were most supportive of CPS processes, particularly in relation to transformation from textual concepts to visual prototypes (RQ2, H2).

3.5.3. Reflective Journals

Throughout the semester, students maintained reflective journals documenting their creative struggles, breakthroughs, and experiences with the CPS-AI framework. To ensure a rigorous evaluation of these qualitative data, the journals were analyzed using thematic analysis [47]. This process involved a deductive coding scheme focused on three primary dimensions:

• Cognitive Engagement: evidence of applying specific CPS stages (e.g., problem restatement, criteria-based selection) to narrative tasks.
• Affective Shift: descriptions of emotional regulation and persistence in overcoming creative bottlenecks or “blank page” anxiety.
• Technical Mediation: reflections on the utility and limitations of specific GenAI tools in the story development process.

Triangulation occurred by cross-validating these qualitative themes with the corresponding numerical shifts in the CPS self-assessment scales. For instance, if a student’s journal entries reflected “increased clarity in story logic through AI-assisted brainstorming,” we verified whether their “Narrative Coherence” score on the quantitative scale showed a synchronized improvement. This mixed-methods approach provided granular insights into RQ2 and RQ3, ensuring that the findings were grounded in both perceived cognitive shifts and documented creative processes.

3.5.4. Expert Evaluation of Final Prototypes (CAT)

The creative quality of the final story prototypes was evaluated using the Consensual Assessment Technique (CAT), the “gold standard” for creativity assessment in domain-specific tasks [32,48]. Three industry professionals, each with more than 15 years of experience in animation production and storytelling, served as independent judges. To ensure objectivity and mitigate researcher bias, a double-blind procedure was strictly enforced: the experts were provided with anonymized versions of the projects without student identities, and they conducted their ratings independently without knowledge of the research hypotheses or the instructor’s expectations [49].

To achieve evaluative consistency while preserving the holistic nature of CAT, a calibration session was held prior to the formal assessment. During this session, the judges were provided with descriptive anchor points for the four evaluation criteria: Originality, Narrative Coherence, Emotional Impact, and Relevance to Audience Trends. For instance, “Narrative Coherence” was anchored by specific indicators such as logical causality and character motivation stability.

Following the orientation, each expert independently rated the prototypes on a seven-point Likert scale for each criterion. Inter-rater agreement was examined using Intraclass Correlation Coefficients (ICC). The resulting ICC values indicated high levels of consistency among the judges, thereby confirming the reliability of the CAT-based evaluations. These objective external ratings were triangulated with student self-assessments to address RQ3 and test H3 regarding improvements in students’ creative performance.

3.6. Data Analysis

Quantitative data from the pre- and post-test CPS scales were analyzed using paired-sample t-tests to examine changes in narrative cognition and affective attitudes (RQ1). Descriptive statistics and one-sample t-tests were used to process the expert CAT ratings and assignment self-assessments (RQ2 and RQ3). All statistical analyses were performed using SPSS v.26, with the significance level set at p < 0.05.

Qualitative data from reflective journals were analyzed following Braun and Clarke’s [47] six-phase thematic analysis framework. To ensure methodological rigor, we employed a hybrid coding approach in which initial codes were deductively derived from the CPS stages and sustainability-related competencies [4,5,6], while sub-themes emerged inductively from the students’ lived experiences with GenAI. To verify the reliability and credibility of the findings, two researchers independently coded a subset (20%). The inter-coder reliability (ICR) analysis yielding a Cohen’s kappa (κ) of 0.84, indicating substantial agreement. Discrepancies were resolved through peer debriefing until a consensus was reached on the final thematic map. The relationship between the initial codes, sub-themes, and final overarching themes is illustrated in

Table 4. Qualitative Coding Scheme and Thematic Development.

Main Theme	Sub-Themes	Initial Codes (Deductive/Inductive)	Sample Student Reflection
Intentional Co-creation	From Prompt-dependency to Narrative Agency	[Crit-AI], [Self-Reflect], [Goal-Align]	“Initially I just copied what AI gave me, but through CPS I learned to treat AI as a mirror to refine my own story logic.”
Affective Resilience	Navigating Creative Blockages	[Persistence], [Frustration-Mgmt]	“When the AI output was ‘hallucinating’, I used the ‘Develop’ phase to debug the narrative rather than giving up.”
Sustainability Reflection	Ethical Judgment in Storytelling	[SDG-Context], [Value-Judgment]	“I used the ‘Clarify’ phase to ensure the ‘Tiger Aunt’ story truly reflected cultural heritage (SDG 11.4).”

.

Finally, the qualitative themes were triangulated with the quantitative findings and expert CAT evaluations. This integration allowed for a richer, multi-dimensional understanding of how the CPS- and AI-supported teaching model influenced students’ learning processes and creative outcomes (RQ2, RQ3), ensuring that the final conclusions were grounded in an evidence chain of both perceived shifts and tangible performance.

During the preparation of this study and manuscript, AI tools, including Perplexity Pro, Gemini (Deep Research), and ChatGPT-5.3 Instant, were used only for limited lan-guage editing, literature screening, and drafting assistance. All substantive decisions re-garding study design, data collection, analysis, interpretation, and final manuscript con-tent were made solely by the author, who carefully reviewed and edited all AI-assisted outputs and takes full responsibility for the content of this publication.

4. Results

4.1. Perceptions of Course Instruction and Learning Environment (RQ1)

Students’ overall perceptions of the CPS- and AI-supported course redesign were positive. Institutional teaching evaluation scores increased from a pre-intervention average of 4.11 to 4.39 during the intervention semester (five-point scale), indicating improved perceived instructional quality and classroom climate.

Qualitative feedback from open-ended comments and reflective journals revealed that structured CPS scaffolding, transparent unit organization, and informal peer discussion channels contributed to a psychologically safe learning environment. To ensure a systematic interpretation of these qualitative insights, we employed a structured coding process as detailed in

Table 5. Qualitative Coding Scheme and Thematic Derivation.

Initial Codes (Open Coding)	Categories (Axial Coding)	Overarching Themes
“Restating the problem,” “Defining constraints,” “SDG goal alignment.”	Problem reframing	Cognitive scaffolding via CPS
“Overcoming fear of starting,” “AI as a partner,” “Persistence after failure.”	Affective resilience	Psychological empowerment
“Prompt engineering,” “Visualizing metaphors,” “Iterative refining.”	Technical mediation	Human-AI co-creativity

. This coding scheme illustrates how raw student reflections were categorized into overarching themes that characterize the learning experience. The semi-structured interview protocol guiding these reflections is provided in Appendix C.

As shown in Table 5, students reported reduced anxiety when sharing preliminary ideas and greater willingness to engage in iterative revision, categorized under the theme of Psychological Empowerment. Furthermore, the integration of GenAI was perceived not as a replacement for effort, but as a form of Human-AI Co-creativity that facilitated the visualization of complex narrative metaphors. Taken together, these findings suggest that the CPS-based design supported not only cognitive development but also affective dimensions of learning, providing a foundational qualitative context for the quantitative shifts observed in RQ1.

4.2. CPS-Related Engagement Across Assignments (RQ2; H2)

Students’ self-assessments across the six assignments indicated consistently high perceived engagement in CPS processes. Mean scores on all five CPS dimensions (fact finding, problem finding, idea finding, solution finding, acceptance finding) exceeded 6.0 on the seven-point scale. One-sample t-tests comparing assignment means to the neutral midpoint (4.0) revealed statistically significant differences for all dimensions across all assignments (p < 0.001), indicating that students perceived each task as strongly supportive of CPS engagement (

Table 6. Self-assessment scores across CPS dimensions (N = 60).

Assignment (Core Task)	Fact Finding (M/SD)	Problem Finding (M/SD)	Idea Finding (M/SD)	Solution Finding (M/SD)	Acceptance Finding (M/SD)	(p)
Assignment 1 Character conflict	6.13/ 0.72	6.18/ 0.65	6.28/ 0.58	6.28/ 0.60	6.18/ 0.70	<0.001
Assignment 2 Suspense imitation	6.10/ 0.75	6.23/ 0.68	6.20/ 0.62	6.20/ 0.65	6.18/ 0.72	<0.001
Assignment 3 Text–to–image	6.23/ 0.68	6.23/ 0.65	6.23/ 0.60	6.28/ 0.55	6.28/ 0.58	<0.001
Assignment 4 “Tiger Aunt” rewriting	6.10/ 0.78	6.15/ 0.72	6.05/ 0.81	6.08/ 0.75	6.10/ 0.78	<0.001
Assignment 5 “Three Regrets”	6.08/ 0.82	6.08/ 0.78	6.15/ 0.75	6.08/ 0.72	6.08/ 0.80	<0.001
Final project (Assignment 6): Story prototype	6.40/ 0.55	6.33/ 0.58	6.38/ 0.52	6.38/ 0.50	6.33/ 0.60	<0.001

Note. One–sample t tests were conducted against the neutral value of 4.0 on the seven–point scale. M = mean; SD = standard deviation.

). The full list of items for the assignment-based CPS self-assessment scale corresponding to Table 6 is provided in Appendix A.

The text-to-image assignment demonstrated particularly strong ratings in idea finding and solution finding. Students reported that transforming textual prompts into visual sequences—supported by optional AI-generated sketches—helped reduce cognitive fixation and facilitated divergent thinking. The personal narrative assignment similarly contributed to perceived authenticity and emotional coherence in storytelling, as students connected their own experiences with character conflict and resolution.

These findings support H2, suggesting that structured resemiotization within CPS scaffolding enhanced perceived transformation efficiency while managing cognitive load, thereby addressing RQ2 regarding students’ experiences of moving from textual concepts to visual story prototypes.

4.3. Pre–Post Changes in CPS-Related Cognition and Affect (RQ1; H1)

Paired-sample t-tests revealed significant improvements in both narrative cognition and affective attitudes from pre-test to post-test. For narrative cognition, mean scores increased from 2.31 (SD = 1.05) to 4.02 (SD = 0.71), t(59) = −12.57, p < 0.001. For affective attitudes, scores increased from 2.44 (SD = 1.08) to 4.09 (SD = 0.75), t(59) = −11.45, p < 0.001 (

Table 7. Pre–post differences in CPS-related competencies (N = 60).

Dimension	Pre–Test (M/SD)	Post–Test (M/SD)	t Value	Significance (p)
Narrative cognition	2.31/1.05	4.02/0.71	−12.57	<0.001
Affective attitude	2.44/1.08	4.09/0.75	−11.45	<0.001

Note. Paired–samples t tests were conducted on a five-point Likert scale.

). The full set of items for the pre–post CPS-related cognition and affect self-assessment scale is provided in Appendix B.

Item-level analysis revealed notable improvements in persistence during complex problem solving and emotional regulation when encountering creative setbacks. Reflective journals corroborated these quantitative findings: over time, students increasingly described iterative revision as a productive process rather than evidence of failure, indicating growth in metacognitive regulation and resilience—competencies closely aligned with sustainability-oriented education frameworks [4,5].

Overall, these results strongly support H1 and address RQ1, suggesting that the CPS- and AI-supported teaching model effectively enhanced both students’ narrative cognition and their affective orientations toward creative problem solving.

4.4. Creative Performance and Expert Evaluation (RQ2; RQ3; H3)

Expert evaluations using the Consensual Assessment Technique indicated substantial growth in creative performance. Across prototypes, experts assigned high ratings in originality, narrative coherence, emotional impact, and social relevance, suggesting that the final story outputs met professional expectations on multiple creativity criteria. Inter-rater reliability (intraclass correlation coefficients) demonstrated acceptable to high agreement, supporting the reliability of the CAT-based evaluations [32].

In the final story prototype, students produced a short animated story integrating the full CPS cycle from problem definition to solution evaluation. To ensure thematic depth, they were required to anchor their narratives in contemporary socio-cultural dilemmas. Examples of student work included: (1) a narrative exploring female agency and AI ethics (“Kidnapped by an AI Boyfriend”), reflecting SDG 5 (Gender Equality); (2) a story on the tension between digital connectivity and physical isolation in urban spaces (“Game Break”), addressing SDG 11 (Sustainable Cities and Communities); and (3) a semi-autobiographical account of the “gig economy” pressures faced by working students (“The Deadline Adventure”), echoing SDG 8 (Decent Work). These prototypes integrated CPS phases from problem clarification and idea generation to solution development and evaluation, demonstrating how narrative cognition was applied to real-world sustainability challenges.

Experts observed that many projects demonstrated sensitivity to authentic social experiences and character-driven transformation rather than reliance on formulaic narrative templates. Qualitative comments emphasized improved alignment between character arcs and thematic messages, stronger emotional logic across narrative sequences, and increased narrative experimentation supported by CPS scaffolding. These findings support H3 and indicate that the CPS- and AI-supported intervention contributed to measurable improvements in creative output quality, thereby addressing RQ2 and RQ3 concerning transformation efficiency and creative performance.

4.5. Model Refinement and Pedagogical Implications

Triangulation of quantitative results, qualitative reflections, and expert feedback identified several areas for refinement. First, while the six-step scaffold effectively supported progressive complexity, stronger explicit integration between suspense-focused and personal narrative tasks may further enhance coherence in character arcs and thematic development. Second, some projects revealed limitations in spatial–temporal plausibility and production feasibility, suggesting a need for more targeted instruction regarding animation constraints and systems logic.

Third, the iterative nature of CPS occasionally contributed to delays in submission timelines, indicating the importance of structured milestones and motivational supports to balance depth of exploration with practical scheduling demands. Finally, the text-to-image assignment emerged as a particularly effective intervention for reducing cognitive fixation, suggesting broader applicability of resemiotization strategies in creative education beyond animation storytelling [25,35].

Collectively, these findings indicate that the CPS- and AI-supported model contributes not only to improved narrative cognition and creative performance but also to sustainability-related competencies, including adaptive thinking, reflective judgment, and collaborative creativity within higher education contexts [4,5,6].

5. Discussion and Conclusions

5.1. CPS–AI Integration as a Metacognitive Mechanism

The practical integration of CPS and GenAI functions as a metacognitive governor [17] that enhances students’ cognitive control over the generative process. Without the CPS framework, students often succumb to “design fixation” or “passive adoption” of AI outputs. In this study, however, the Clarify phase established ‘evaluative benchmarks’ before any AI interaction, ensuring that students approached the tool with intentionality. During the Ideate and Develop phases, GenAI served as a cognitive perturbation device [31], while the CPS ‘hits and highlights’ technique forced students to exercise high-level cognitive control—critically selecting and refining AI suggestions based on the previously defined thematic goals. This integration effectively shifted the student’s role from a “prompt consumer” to a “narrative architect” and provided the structural discipline needed to navigate AI’s vast visual possibility space without losing narrative coherence.

5.2. CPS as a Cognitive Bridge for Heterogeneous Learners (RQ1)

In relation to RQ1, a central contribution of this study lies in demonstrating that a CPS- and AI-supported teaching model can function as a cognitive bridge for heterogeneous learners in higher education. Although many participants possessed technical animation skills, they initially struggled with narrative problem framing, emotional coherence, and structured story development, reflecting what creativity research identifies as deficiencies in problem finding—a foundational yet often underdeveloped component of creative competence [8,9].

By explicitly mapping CPS stages onto instructional units and assignments, the intervention provided a shared cognitive structure through which students could organize fragmented experiences into systematic narrative reasoning processes. This structure appeared particularly beneficial in a mixed cohort composed of design, business, and interdisciplinary students, who were able to use CPS terminology and steps as a common language for collaboration. While natural student maturation or peer collaboration (social learning) undoubtedly contributed to skill development, the specific patterns of improvement—particularly the significant gains in Problem Framing and Visual Resemiotization—align precisely with the targeted CPS-AI scaffolds provided in Units 3 and 4. This temporal correspondence suggests that the pedagogical model, rather than generic academic progress, was the primary driver of the observed cognitive shifts.

The findings resonate with componential theories of creativity, which emphasize the interaction between domain-relevant skills, creativity-relevant processes, and motivation [15]. In this context, CPS served as a process-level scaffold that enabled students with diverse strengths to collaborate productively: visually oriented students contributed to ideation and character design, while analytically inclined students strengthened feasibility assessment and logical structuring. From a sustainability education perspective, such interdisciplinary collaboration reflects the interpersonal and integrative competencies required for addressing complex societal challenges [4,5]. Thus, CPS did not merely enhance storytelling skills; it fostered collaborative problem-solving capacities relevant to sustainability-oriented higher education.

5.3. Transformative Mechanisms in CPS- and AI-Supported Learning (RQ2; RQ3)

The significant improvements observed in students’ narrative cognition and affective resilience (p < 0.001) merit a critical discussion regarding their attribution. While maturation effects are a common threat to internal validity in single-group designs, the statistical magnitude and rapid nature of these gains within a compressed 12-week timeframe suggest an effect that exceeds typical longitudinal maturation. These findings align with the observations of Li et al. [50], who argued that GenAI acts as a potent catalyst for ideation by lowering the threshold for visual prototyping.

The empirical results from the CAT expert evaluations (ICC > 0.80) directly validate this growth in creative performance. Specifically, the significant jump in Narrative Logic (refer to Table 7: M_pre = 2.31 to M_post = 4.02) and Affective Resilience (from M_pre = 2.44 to M_post = 4.09) correlates with the qualitative evidence found in student reflective journals, where participants described moving beyond passive “AI-hallucination” toward “intentional co-creation.” This alignment between independent expert ratings and student reflections provides a robust evidence chain, demonstrating that the observed gains were driven by the structured CPS-AI navigation rather than generic academic progress.

Building on this empirical foundation, the CPS-AI teaching model appeared to reconfigure how novice storytellers approached uncertainty and iterative refinement. Reflective journals revealed that CPS prompts (e.g., problem restatement, criteria-based selection) helped students reinterpret revisions as a productive part of development. At the same time, GenAI tools were described as useful catalysts for breaking through “blank page anxiety.” However, our results diverge from the “homogenization” risk identified in recent creative writing studies [40], which cautioned that AI assistance often leads to a reduction in collective diversity. In contrast, by anchoring the generative process within a structured CPS “navigation system,” our students maintained high levels of Originality.

This divergence suggests that the “scaffolded perturbation” provided by CPS functions as a necessary counterbalance to the convergence bias inherent in Large Language Models (LLMs). Within our course design, CPS explicitly required learners to articulate selection criteria (e.g., thematic coherence, social relevance) and reject AI outputs when they conflicted with narrative goals. This process aligns with sustainability competency frameworks [4,5,6] by merging anticipatory thinking with critical assessment. Consequently, GenAI operated not as an unqualified accelerator, but as a conditional amplifier whose educational value depended on the CPS framework cognitive scaffolding. This integration offers a scientifically grounded solution to the “efficiency-depth paradox,” ensuring that technical acceleration does not necessitate a loss of narrative complexity or creative agency.

5.4. Theoretical and Practical Implications: Bridging CPS, AI, and Sustainability

The unique contribution of this study lies in its empirical illustration of a “co-evolutionary navigation” model—a conceptual shift from viewing GenAI as a mere production tool to treating it as a partner within a structured cognitive loop. At the theoretical level, the findings extend scaffolded learning perspectives into the era of generative AI, suggesting that “scaffolds” need not be exclusively human-led but can be process-led through the intentional orchestration of CPS stages and algorithmic stimuli. In this sense, CPS provides the metacognitive structure for problem framing and evaluation, while GenAI supplies adjustable representational stimuli within that structure. For design educators struggling with the “efficiency–depth paradox” in post-AI classrooms, this model offers a concrete way to accelerate visual exploration without sacrificing narrative coherence or critical judgment.

From a sustainability education perspective, a key contribution of this study lies in demonstrating how abstract sustainability-related competencies can be translated into concrete creative media practices. The CPS- and GenAI-supported model was intentionally designed to align specific CPS phases with recognized sustainability-related competencies [4,5]. Rather than treating sustainability education as limited to explicitly environmental content, the intervention frames narrative transformation as a core mechanism for social and cultural sustainability. For instance, by rewriting “Tiger Aunt” (Assignment 4), students engaged with SDG 11.4 by exercising the normative competence required to challenge outdated social fears and reimagine cultural heritage. Similarly, the “Three Regrets” (Assignment 5) addressed sustainability at the micro-level—family and self—by using narrative resolution to foster affective resilience (SDG 3). As emphasized by UNESCO [6], sustainability requires the social and emotional intelligence to navigate conflicts and preserve cultural identities; tasks that ask students to articulate moral tensions and revisit decisions through iterative refinement draw directly on normative, strategic, and interpersonal competencies.

The qualitative characteristics of the final story prototypes further illustrate how these competencies can be rehearsed within an animation storytelling context. In “Kidnapped by an AI Boyfriend,” students explored tensions between female agency and AI-mediated control, inviting critical reflection on power, consent, and technological dependence. This kind of value-laden narrative engages normative competence and anticipatory thinking, as learners imagine the longer-term implications of everyday interactions with intelligent systems [4,5,6]. “Game Break” addressed the tension between digital hyper-connectivity and physical isolation in urban environments, foregrounding questions of livability, public space, and community belonging; by constructing alternative futures for their characters, students practiced systems-thinking and anticipatory competencies, considering how individual choices unfold within wider socio-technical systems. “The Deadline Adventure,” a semi-autobiographical account of gig-economy pressures faced by working students, required learners to articulate trade-offs between economic survival, mental health, and educational aspirations, thereby exercising strategic and interpersonal competencies as they imagined more sustainable ways of organizing work and study [4,5,6]. Collectively, these prototypes suggest that the CPS- and GenAI-supported model did not merely increase narrative sophistication in a generic sense, but also created conditions for students to rehearse sustainability-related ways of thinking—such as evaluating value conflicts, anticipating systemic consequences, and negotiating social relationships—within a familiar yet challenging medium.

These findings resonate with existing frameworks that define sustainability-related competencies as integrated cognitive and affective capacities [5,6]. However, while recent reviews emphasize that many higher education initiatives struggle to make these competencies tangible, this study provides a substantive response to this gap by demonstrating how value-laden socio-cultural dilemmas can be tackled through multimodal, AI-supported resemiotization. By coupling CPS scaffolding with GenAI, we illustrate a concrete pathway for integrating sustainability into design curricula. Ultimately, the observed gains in narrative cognition and affective resilience suggest that creative media courses can function as transformative learning environments, offering a model for harnessing AI’s benefits while maintaining the depth required for sustainability-oriented problem solving.

Specifically, the participants were students in a digital multimedia design department within a vocational education system. Their prior technical skills (e.g., proficiency in animation software) and vocational learning orientations may differ from those of students in liberal arts or research-intensive universities. To address this, we have provided a “thick description” [51] of the instructional context and student demographics, allowing other educators to assess the “analytical transferability” of the CPS-AI model to their own specific settings. These implications should be interpreted in light of several methodological and contextual limitations, which are discussed in detail in Section 6.1.

5.5. Conclusions

This study demonstrates that integrating Creative Problem Solving and generative AI within animation storytelling education can enhance narrative cognition, affective resilience, and creative performance while fostering sustainability-related competencies. The CPS- and GenAI-supported model functioned not only as a pedagogical innovation in media education but also as a competency-development framework aligned with sustainability-oriented higher education. Quantitative gains, expert evaluations, and qualitative reflections collectively indicate that structured creative processes can cultivate adaptive thinking, reflective agency, and collaborative problem-solving capacities.

As higher education institutions seek scalable approaches to embed sustainability-related competencies across disciplines, this study offers an empirically grounded model illustrating how creative media practice can serve as a site for transformative learning [4,6]. Future research may explore longitudinal impacts, cross-institutional replication, and deeper integration of sustainability themes within narrative content itself, as well as comparative studies across different creative disciplines. Nonetheless, the present findings suggest that CPS- and AI-supported storytelling education holds significant promise for advancing sustainability-related competencies in higher education.

6. Limitations and Future Directions

6.1. Limitations

Several limitations should be acknowledged when interpreting the findings of this study. First, although the sample size (N = 60) is consistent with classroom-based intervention research, participants were drawn from a single university of technology in Taiwan. Institutional culture, disciplinary composition, and regional educational norms may have influenced both implementation and outcomes. Consequently, generalizability to other institutional types—such as comprehensive universities, liberal arts colleges, or international programs—remains to be empirically tested.

Second, the one-group pre–post design limits causal inference. Without a control group, it is not possible to fully isolate the effects of the CPS- and AI-supported intervention from alternative explanations, including maturation effects, peer collaboration dynamics, or concurrent coursework. Although triangulation through expert CAT evaluations and qualitative reflections strengthens internal validity, future research employing quasi-experimental or randomized controlled designs would allow for stronger causal claims.

Third, quantitative data relied substantially on self-report instruments. While such measures are appropriate for assessing perceived competence and affective change, they are susceptible to social desirability bias and may overestimate actual performance gains. Although expert evaluations partially addressed this concern, incorporating independent blind ratings, rubric-based performance assessments, or longitudinal portfolio analysis would provide more objective validation.

Fourth, the instructor simultaneously served as the primary researcher. Despite efforts to mitigate bias through structured rubrics, expert panels, and transparent procedures, this dual role may have influenced implementation fidelity and interpretation. Future studies could involve independent instructors or multi-site collaboration to enhance methodological neutrality.

Finally, the absence of longitudinal follow-up limits understanding of the durability of observed gains. sustainability-related competencies—such as systems thinking, strategic reasoning, and reflective agency—are developmental in nature [4,5]. Without delayed post-tests or subsequent performance tracking, it remains unclear whether the improvements observed during the course persist, transfer, or evolve over time.

Overall, these limitations suggest that while the present findings offer promising evidence for a CPS- and AI-supported storytelling model, they should be interpreted as an initial contribution that requires further corroboration across designs, contexts, and timescales.

6.2. Future Research Directions

Building on these limitations, several directions for future inquiry emerge.

First, replication and validation of the CPS-AI framework across diverse cultural and institutional landscapes are essential. While the current results are promising, future studies should include different geographic regions and disciplinary backgrounds—such as comprehensive universities or international liberal arts programs—to establish the model’s broader applicability in global design education. Comparative multi-site studies could further clarify how institutional culture, regional educational norms, and technological infrastructure influence implementation outcomes in sustainability-oriented higher education [6].

Second, stronger experimental designs should be pursued where feasible. Quasi-experimental comparisons, matched cohort studies, or randomized controlled trials would enhance the evidence base for CPS- and AI-supported pedagogy and allow more precise estimation of its effects relative to traditional instruction. To further enhance methodological neutrality, future studies should employ independent instructors who are not involved in the research design, or utilize multi-site collaborative teams. This would allow for a clearer separation between pedagogical execution and data analysis, further reducing the risk of experimenter bias.

Third, interdisciplinary comparative research may reveal how students from distinct domains—such as engineering, business, humanities, or environmental studies—engage differently with CPS-based storytelling tasks. Such insights could inform differentiated scaffolding strategies and support integration competence across disciplinary boundaries, responding to calls for more cross-cutting approaches to sustainability-related competencies [4,5].

Fourth, future studies may incorporate multimodal and process-oriented data collection methods. Learning analytics, process logs, collaborative discourse analysis, and cognitive load indicators could deepen understanding of how resemiotization strategies and GenAI tools shape creative cognition and group dynamics. Such data would help clarify the mechanisms through which AI functions as a cognitive collaborator rather than a replacement for human agency [31].

Fifth, expanding the substantive integration of sustainability themes represents a promising avenue. While the present study focused primarily on competency development, future iterations could explicitly embed socio-ecological challenges—such as climate adaptation, urban resilience, or social equity—into narrative assignments. This would allow students to apply CPS processes to real-world sustainability dilemmas, thereby strengthening anticipatory and strategic competencies [5,6].

Finally, there is potential to establish cross-institutional repositories of CPS-based curricula, AI-guided assignment frameworks, and validated evaluation rubrics. Such collaborative infrastructures could support iterative refinement, scalability, and policy-level integration of sustainability-related competencies within creative media education, and facilitate knowledge sharing among educators designing GenAI-supported creative courses.