From tools to thinking partners: Cognitive and pedagogical shifts in design education through generative AI

Abstract

Generative Artificial Intelligence is reshaping design education by influencing how students develop ideas and make creative decisions. Unlike earlier digital shifts that mainly changed production methods, generative AI introduces cognitive changes by moving part of ideation and evaluation into human–AI collaboration. This study offers a rare perspective by comparing design educators’ views on integrating AI in El Salvador and Indonesia, with Denmark as a comparison case. Using a qualitative interpretive approach, interviews with nine academics reveal key themes such as AI as a thinking partner, the persistence of manual-first pedagogy and the rising importance of prompt literacy. Findings show that in resource-limited programs, manual-first traditions provide an important foundation for making curatorial judgments about AI-generated ideas, while Denmark educators use a more structured integration aligned with industry demands. Design educators emphasise ethical awareness and critical judgment, framing AI as a catalyst for rethinking creative processes rather than replacing them.

Introduction

Technological change has always been an integral part of the evolution of design and its educational practices. Each major technological innovation, from the introduction of the Macintosh computer and digital design software in the 1980s to the spread of digital photography, interactive media, and web design in the late 1990s, has redefined the creative dimension of design practice (Armstrong and Stojmirovic, 2011; Drucker and McVarish, 2013). These technological transformations required design educators to continuously revise curricula, pedagogies, and assessment strategies (Fleischmann, 2013). The transition from analogue production to digital tools reshaped how designers worked, while a later shift toward interactive media prompted the emergence of new fields such as user interface (UI) and user experience design (UX). Yet, despite the magnitude of these changes, design creativity remained a fundamentally human-centric practice. Generative Artificial Intelligence (AI) shifts this human-centric design process to computer-generated automation that can bypass human ideation and iteration central to design creativity (Fleischmann, 2025a; Wadinambiarachchi et al., 2024).

Unlike earlier technological shifts that changed how designers executed ideas, generative AI produces new material from user input known as prompts, which are text-based instructions that communicate design intent to the system. Text-to-image systems such as Midjourney, DALL-E 2, and Adobe Firefly, as well as large language models such as ChatGPT, can now generate ideas, variations, and explanations that were once central to design students’ own creative reasoning and early design critique. In doing so, the use of these generative AI tools invites fundamental questions about authorship, originality, and cognition in design. As Gao et al. (2025) note, AI now supports all stages of the design process, from inspiration to evaluation, effectively acting as a “design material”. Fleischmann (2025b) views this shift as the emergence of AI as a “co-designer,” arguing that generative AI systems not only extend human ideation but actively participate in creative decision-making, which requires educators and students to rethink how authorship and creativity are negotiated in design practice.

In this article, generative AI is understood as an algorithmically trained system that participates in ideation and evaluation through human–AI interaction; prompt literacy refers to the ability to articulate design intent through language in order to guide AI outputs; and curatorial judgement describes the designer’s role in selecting, refining, and rejecting AI-generated material.

Within this conceptual framing, generative AI developments bring into question the boundaries between human and computational creativity in design education. The literature increasingly emphasises that design educators must help students develop skills such as prompt engineering, critical reflection, and ethical awareness for effective use of AI (Fleischmann, 2025a; Lee, 2025; Muji et al., 2023). Researchers describe AI as part of a distributed cognitive system in which aspects of thinking and memory move from the human mind to external systems (Clark and Chalmers, 1998; Smart et al., 2025). Within design education, this redistribution of the creative process raises pressing questions: What remains tacit or original when AI externalises part of ideation? How should design educators cultivate reflection and judgement in a context where generative systems supply endless visual and textual solutions?

While recent research on generative AI in design education has grown rapidly, much of it focuses on well-resourced Western institutions. Studies from the United States, China, and Australia (Sitthiworachart et al., 2024) explore how AI enhances creativity and efficiency, but these studies rarely consider institutions with limited technological infrastructure or different pedagogical traditions. This imbalance risks constructing a narrow view of generative AI’s educational potential across global design programs.

To address this research imbalance, this study investigates how design academics in El Salvador, Indonesia, and Denmark (interviewed in Australia) perceive the influence of generative AI on teaching and learning. The perspectives from El Salvador and Indonesia are particularly significant because they originate outside the dominant centres of AI research and design education. Including Denmark provides a point of comparison with a well-resourced European system, enabling a more nuanced analysis of how institutional conditions shape design educators’ approaches to AI. Together, these cases offer a comparative lens for examining how generative AI reshapes cognitive, pedagogical, and ethical dimensions of design across higher education design programs that differ in resources, culture, and institutional infrastructure.

Literature review

Technological transitions in design education: From analogue to digital

The history of design education reflects a continuous process of adaptation to technological change. The transition from analogue to digital production in the 1980s introduced an unprecedented reconfiguration of visual communication workflows (Drucker and McVarish, 2013). The release of the Apple Macintosh in 1984, together with PostScript and the layout software PageMaker, enabled the rise of desktop publishing, transforming how designers conceptualised and produced visual artefacts. Design educators were compelled to integrate new forms of digital literacy into their curricula, replacing mechanical reproduction skills with competencies in page layout, typography, and image manipulation using software such as Adobe Illustrator and Photoshop.

This first wave of digitisation established a model of curricular response that emphasised technical tool mastery. Training design students to use computers efficiently was regarded as essential for employability and professional relevance in both design education (Gu et al., 2010) and broader educational contexts (Howard and Mozejko, 2015). Yet, as researchers later observed, these curricular adjustments centred largely on adopting new tools rather than rethinking the cognitive or creative foundations of design pedagogy (Loh, 2017). The reflective and experiential foundations of studio education largely remained intact; what changed was the medium through which these pedagogies operated.

By the late 1990s and early 2000s, the internet and interactive media triggered another pedagogical shift. As design extended into digital interfaces, websites, and mobile platforms, new professions such as interaction designer and user-experience designer emerged (Nielsen, 2017). These technological transitions profoundly affected what and how design was taught but they did not alter the fundamental purpose of design education, the development of a reflective mindset (Schön, 1983). Design educators adapted by layering new digital skills onto existing pedagogical frameworks instead of reconstructing the foundations of design cognition. This pattern provides an important historical benchmark for interpreting the arrival of generative AI.

From digital media to generative AI: A qualitative shift

Generative AI introduces a more radical disruption because it engages directly with ideation and reasoning, the core of design cognition. While digital technologies extended human capabilities, generative systems externalise aspects of imagination and evaluation once considered uniquely human (Hou et al., 2025; Zhou and Lee, 2024). Unlike previous waves of digitisation, which required mastering new interfaces, generative AI requires new forms of thinking shaped by the interplay between human intention and algorithmic output.

Design education researchers increasingly describe AI not merely as a production tool but as a co-creative collaborator (Fleischmann, 2025b; Gökdemir and Kayan, 2025). For instance, Gökdemir and Kayan (2025) conceptualise AI as a “linguistic interface” that enables designers to externalise ideas through conversational prompting. This linguistic turn shifts creative agency from drawing or sketching to articulation and dialogue, altering the sequence in which designers frame problems and generate visual outcomes.

Empirical studies across disciplines confirm that generative AI is entering multiple stages of the design process from early ideation and research to prototyping and presentation (Buendía-García, 2025; Fleischmann, 2024, 2025a). This breadth of application indicates that AI is dissolving the distinction between conceptual and technical phases of design. Fleischmann (2025b) argues that the rapid accessibility of AI imagery risks replacing the reflective space traditionally afforded by sketching and iteration, potentially undermining tacit knowledge and experiential learning. Nevertheless, other researchers report increased creative fluency and ideation diversity, suggesting that generative AI expands rather than replaces human creativity (Hwang and Wu, 2025; Lee, 2025).

The notion of a qualitative shift is supported by theoretical frameworks such as distributed cognition and extended mind theory (Clark and Chalmers, 1998). AI functions as an external cognitive system that stores, retrieves, and transforms information collaboratively with the designer. In this sense, design cognition becomes hybrid: part human, part algorithmic. As Gökdemir and Kayan (2025) argue, AI now functions as a cognitive co-agent that reshapes how design students think and make decisions, demanding pedagogical frameworks that cultivate critical evaluation and reflective judgement of algorithmic outcomes (see also Fleischmann, 2025a; Rana et al., 2025).

Cognitive and creative transformation

Cognitive psychologists describe creativity as a dynamic interplay between divergent and convergent thinking, mediated by reflection and evaluation (Eymann et al., 2024). Generative models challenge this dynamic by externalising divergent thinking: generative AI tools can create hundreds of visual options instantaneously, effectively outsourcing ideation to computational processes (Rana et al., 2025). As a result, students risk bypassing the cognitive struggle that traditionally drives creative insight (Fleischmann, 2025a).

Studies in design and architecture programs show both gains and trade-offs when using generative AI. In early-stage concept exploration, generative systems can broaden the ideation space and speed visualisation (Kwon et al., 2024; Rana et al., 2025). Bai (2025) found that design students using AI-integrated tools achieved higher creativity scores and faster task completion compared to traditional methods. However, other studies report mixed outcomes. Wadinambiarachchi et al. (2024) observed that generative AI can increase idea variety but also lead to design fixation and loss of stylistic diversity, while Gong et al. (2024) noted that outsourcing ideation to AI risks reducing originality in concept development. These findings echo concerns raised by Fleischmann (2025b) that generative AI may erode the embodied and reflective aspects of studio practice if adopted uncritically.

From a pedagogical standpoint, this transformation can be interpreted through Polanyi’s (1966) distinction between tacit and explicit knowledge. Traditional studio education cultivates tacit understanding, which is knowledge embedded in doing, seeing, and sensing (Fleischmann, 2025b). When AI systems generate outputs based on textual prompts, the locus of learning shifts from making to curating. Students become evaluators of pre-generated alternatives rather than creators of artefacts from first principles. This change introduces a new form of cognitive offloading (Clark and Chalmers, 1998; Shukla et al., 2025), where generative AI performs exploratory functions once carried out by the designer’s own reflective imagination.

However, some researchers interpret this transformation as an expansion of human cognition rather than a loss. Gökdemir and Kayan (2025) argue that interacting with AI’s linguistic and visual outputs stimulates metacognitive reflection, helping students articulate design rationale more clearly. Hwang and Wu (2025) similarly observe that generative AI can encourage students to reassess and clarify their ideas, influencing how they structure creative decisions.

While various design education researchers cast generative AI as a tool that streamlines design workflows (e.g., Bartlett and Camba, 2024), another viewpoint is emerging that generative AI is shifting how designers think. This school of thought argues that generative AI reshapes the cognitive ecology of design learning by redistributing creative functions between human intuition, linguistic articulation, and machine computation. For design educators, this raises important questions about how to balance automation with depth of thinking, understanding and exploration, and the reflective practice that sit at the core of design education.

Pedagogical shifts: Studio learning in the age of generative AI

Design education has long relied on studio-based pedagogy, which emphasises iterative making, critique, and reflection-in-action (Fleischmann, 2025b; Schön, 1983). The introduction of generative AI is transforming this environment by changing how design students engage in experimentation, iteration, and feedback. Studies across multiple design disciplines demonstrate that AI is increasingly integrated into studio projects as a creative partner rather than as a technical instrument.

Tellez (2023) observed that the introduction of text-to-image systems such as Midjourney into undergraduate design studios expanded students’ ideation range and collaborative behaviours, demonstrating that AI can transform not only workflows but also classroom dynamics. Buendía-García (2025) reports that integrating AI into UX-design courses fosters collaboration between students and the generative AI system, prompting discussions about authorship, originality and the aesthetic limits of automation.

AI has begun to reshape early design workflows, with studies showing that generative tools support rapid variation-making and expanded ideation. Tellez (2023) observed that AI accelerates early-stage ideation by enabling students to generate multiple variations quickly, while Gao et al. (2025) show that AI-based form generation and simulation compress production time and broaden the space of conceptual possibilities. This acceleration requires design educators to re-evaluate assessment criteria. When iterations are generated within seconds, the pedagogical focus must shift from quantity of outcomes to quality of reasoning and justification.

The notion of ‘prompt literacy’, which is the ability to articulate design intent through language, has emerged as a critical skill in this context (Fleischmann, 2025a; Gökdemir and Kayan, 2025). Design students must learn to translate conceptual ideas into precise textual or visual cues that elicit meaningful AI responses. This process requires both linguistic skills and critical reflection. Engaging with AI generated outputs further reinforces this reflective dimension, as design students examine how their choices of wording, structure and intent shape the system’s behaviour, supporting the development of meta-cognitive awareness in the design process (Chen et al., 2025).

Ethics, authorship, and epistemic vigilance

AI introduces complex ethical and epistemic questions that extend beyond workflow or pedagogy. Researchers highlight ongoing concerns surrounding authorship, originality and accountability in design education (Amini, 2025; Bartlett and Camba, 2024). When AI outputs are co-generated by algorithms trained on large, often uncurated datasets, the boundaries of creative ownership become less clear. Students may unintentionally reproduce biased or copyright-protected material challenging established definitions of originality and underscoring the need for stronger ethical literacy in studio practice.

Design educators emphasise the importance of building ethical awareness and critical reflection, including the ability to question the accuracy, bias and provenance of AI-generated outputs. This capacity aligns with what cognitive scientists describe as epistemic vigilance (Sperber et al., 2010) which is increasingly embedded in contemporary design critique. Scandinavian programs, for example, integrate discussions of dataset bias, authorship attribution and sustainability into AI-related assignments (Muji et al., 2023). Similarly, Fleischmann (2025b) frames ethical reflection as a creative competency, arguing that questioning the origins and implications of AI outputs must become part of design thinking.

Empirical studies in architecture and product-design education support this shift. Hwang and Wu (2025) and Huh et al. (2025) show that while students value AI’s efficiency, they also question the authenticity of AI-generated outcomes and feel less emotionally connected to work produced through automated processes. These concerns highlight that ethical literacy cannot be reduced to technical proficiency; it also requires critical judgement and the ability to evaluate the meaning, origin and quality of AI generated outputs. In other words, students need to learn not only how to use AI but also when and why to trust it.

Context matters: Access, localisation, and global perspectives

Most research on AI in design education originates in well-resourced settings in North America, Australia, Europe and parts of East Asia (Sitthiworachart et al., 2024). Emerging studies from Latin America indicate that when access to technology is uneven, educational innovation often stems from adaptation to local realities rather than the mere availability of advanced tools. For example, universities in Chile and Mexico have integrated AI through strategies that combine curriculum redesign, teacher training and community partnerships, including projects to preserve Indigenous languages and knowledge systems (Barrera and Azeez, 2024). While this example addresses higher education broadly rather than design education specifically, it illustrates how resource constraints shape institutional approaches. Similarly, a recent undergraduate thesis from Colombia highlights efforts to update graphic design education by incorporating AI tools while addressing infrastructure limitations and the need for teacher capacity building (Sánchez Jaime et al., 2023). Regional policy studies underscore that connectivity gaps and resource constraints influence implementation priorities, often emphasizing ethical frameworks and inclusive practices over rapid technological adoption (ProFuturo and OEI, 2024). Recognising this contextual diversity is essential to avoid a single, Western-centric narrative of AI innovation.

Methodology

Against this backdrop, the present study investigates how design educators in Indonesia and El Salvador, alongside educators teaching in Denmark, understand and integrate generative AI within their curricula. This comparative focus enables an examination of how contrasting educational infrastructures, from resource-constrained public universities to digitally advanced European programs, shape the pedagogical and ethical interpretation of emerging technologies. The study explores specifically how design educators in different sociocultural and institutional settings interpret the affordances, challenges and pedagogical implications of generative AI in their teaching. Because the research seeks to capture lived experience and contextual meaning rather than to generalise statistically, a qualitative-interpretive approach underpinned by pragmatism was adopted (Kaushik and Walsh, 2019; Morgan, 2014). Pragmatism positions knowledge as experiential and situational, allowing methods to be chosen that best address practice-based questions within real-world educational settings (Morgan, 2014).

Research design and participants

Semi-structured group and individual interviews were conducted between 2024 and 2025. The sample included eight design educators and one marketing lecturer working with design students across the three countries. Participants were recruited through opportunity-based purposive sampling during periods of international research travel and academic exchange, rather than using a predefined sampling frame. While travelling in Indonesia and El Salvador, the researcher identified local design programs and contacted design educators within those programs using publicly available university contact details to request interviews. The Danish participants were interviewed during a visiting academic exchange at the researcher’s institution.

This pragmatic, opportunity-based approach is appropriate for exploratory qualitative studies that seek context-rich insight into emerging practices, rather than aiming for statistical representativeness (Morgan, 2014; Patton, 2015).

The inclusion of less-resourced design programs allowed exploration of how local educational cultures and funding conditions influence perceptions of generative AI. Participants included lecturers, senior academics, and curriculum coordinators specialising in graphic design, visual communication, multimedia and interaction design, design research, and marketing.

Interview participants, their countries, and their specialities included:

El Salvador: Three academics from the Universidad Tecnológica de El Salvador (UTEC) teaching visual communication and multimedia design.

Indonesia: Four design educators from the Institut Seni Indonesia Yogyakarta and Institut Seni Indonesia Surakarta representing visual communication, interaction design and design research.

Denmark (interviewed in Australia): Two lecturers in graphic design and marketing from the Danish School of Media and Journalism, Copenhagen who were interviewed while visiting the researcher’s university.

All participants had experience integrating or debating generative AI tools within their curriculum. For simplicity, all participants are referred to as design educators in the following discussion. Ethics approval was obtained from Griffith University and participants were fully informed of the study’s aims and gave consent for anonymised use of quotations.

Data collection

Interviews were conducted in English, Spanish, and Indonesian. In El Salvador, interviews were held in Spanish; the researcher, who speaks Spanish, later provided recordings to a professional translator for English transcription and reviewed the translations for clarity. In Indonesia, interviews were conducted either in English or with the assistance of an English-speaking interpreter. Danish interviews were conducted in English. This multilingual approach prioritised participants’ comfort while aiming to preserve conceptual meaning across languages (Temple and Young, 2004).

The interview sessions ranged from 45 to 90 minutes and followed a semi-structured format designed to encourage open reflection (Kvale and Brinkmann, 2015) on teaching practices, perceived changes in how design students think and learn as they use generative AI tools and ethical considerations of AI use. Group interviews in El Salvador and Indonesia enabled collective group reflection typical of design-studio discourse, while the Danish interviews were held individually, as the two participants held different disciplinary roles and separate sessions allowed their perspectives to be examined more fully. Field notes were also used with the researcher recording contextual details and reflections immediately after each interview.

Data analysis

An inductive thematic analysis (Braun and Clarke, 2006; Kiger and Varpio, 2020) was employed to identify patterns across the dataset. Transcripts were organised and coded manually using Microsoft Word and Excel. Initial topic codes captured broad areas such as perceived ‘opportunities’, ‘challenges’, and ‘pedagogical implications’ (Hahn, 2008). These topics were refined through successive cycles of analysis and second-order coding (Saldaña, 2013) to identify emergent sub-themes, including ‘AI as a catalyst for reflection’, ‘ethical uncertainty’, and ‘changing definitions of authorship’. Codes were iteratively compared within and across country datasets to reveal both shared and context-specific perspectives.

To strengthen rigour, initial coding was reviewed by a research assistant not involved in data collection. The researcher also kept reflexive notes to acknowledge her dual role as educator and investigator, helping to monitor potential bias and interpretive influence (Lincoln and Guba, 1985).

While the study did not aim to systematically compare differences within countries, the researcher noted variation linked to institutional resources, disciplinary focus, and individual teaching philosophy. These nuances are referenced in the analysis where relevant.

Presentation of findings

Findings are presented thematically, supported by verbatim quotations labelled by country (ID - Indonesia, ES - El Salvador, DK - Denmark) and participant role and number (e.g. design educator 1 > DE 1. Quotations illustrate individual and collective perspectives and enable readers to see how interpretations were grounded in participants’ voices (Corden and Sainsbury, 2006). Overall, the methodology reflects a pragmatic–interpretive approach, emphasising contextual insight and comparative understanding over generalisation. By juxtaposing perspectives from Indonesia, El Salvador, and Denmark, the study shows how cross-national diversity and local conditions shape the pedagogical negotiation of generative AI in contemporary design education. The aim of this study is to offer insight into emerging discourses surrounding generative AI in regions underrepresented in international design-education research.

Where the analysis refers to industry expectations, institutional policies, or regional conditions, these statements reflect participants’ perceptions and experiences as reported in interviews, unless otherwise supported by external sources.

Limitations

This study has several limitations. With a small sample (n = 9) and opportunity-based purposive recruitment, the findings are not intended to be generalisable across national design education systems. The selection of countries reflects research access rather than a systematically comparative design; cross-national contrasts should therefore be read as indicative rather than definitive. The study adopts an exploratory qualitative approach, offering contextual insight into how generative AI is interpreted by design educators in different institutional settings rather than representative conclusions about global practice.

Findings

AI as tool, collaborator, and disruptor

(identified by 8 participants: ID-DE 1–4, ES-DE 1–3, DK-DE 1)

Across all countries, design educators described AI as a powerful tool that accelerates exploration of ideas yet consistently stressed that human judgement and design intent must remain central to students learning design principles. One participant from El Salvador summarised this viewpoint: “AI is a tool that helps us explore, but the designer must decide what is functional or aesthetically valid” (ES-DE 1). Similarly, Danish interviewees saw generative AI as a “thinking partner” that supports, but does not replace, critical reflection.

In Indonesia, the framing was more ambivalent. Design educators acknowledged AI’s creative potential but feared it could undermine manual craft traditions: “It’s fabulous for experimentation, but in the end you have to be the human who decides what’s right” (ID-DE 2). The term “companion tool” appeared repeatedly, reflecting a cautious acceptance of generative AI among Indonesian design educators.

Despite local nuances, the dominant narrative positioned AI as augmentative rather than autonomous tool, echoing broader global discourse that creativity remains a human-led process.

Manual-first pedagogy as cognitive anchor

(identified by 5 participants: ID-DE 1-4, ES-DE 2)

Indonesian design educators repeatedly described manual design processes like drawing, collage, and model-making as cognitive grounding for developing sensitivity and aesthetic judgement. “The manual process builds the sensitivity that’s essential… later, technology becomes easier to master” (ID-DE 1). Another noted: “AI can be used for inspiration, but the final work must be manual [in the design foundation year]” (ID-DE 3). This manual-first stance is both pedagogical and cultural, representing resistance to reliance on algorithmic automation and an assertion of local identity within a globalised profession. El Salvadoran participants shared similar sentiments, framing manual work as a form of “discipline” that maintains reflective depth.

This finding suggests that in less-digitally saturated countries, manual and material forms of learning remain important for cultivating critical awareness and providing a counterbalance to AI-driven efficiency.

From making to directing: Changing creative processes

(identified by 6 participants: ID-DE 1, 2 and 4, DK-DE 1–2, ES-DE 1)

A major theme expressed by the design educators was the perceived shift in the designer’s role from maker to director or curator. Danish participants emphasised that students must now articulate conceptual intent more explicitly to guide AI: “They become directors rather than designers, which changes the creative process” (DK-DE 1).

In Indonesia, one design educator echoed this perspective: “They [students] have to justify every prompt. AI becomes part of their process, not a shortcut” (ID-DE 4). Similarly, an El Salvadoran design educator emphasised the need to retain reflective, process-based craftsmanship, cautioning: “If we use it just to get quick results, we lose the process… design requires reflection and judgement” (ES-DE 1).

This shift from maker to director or curator reframes creativity by placing greater emphasis on conceptual thinking and critical judgment, while manual production skills remain an essential foundation in Indonesia and El Salvador. It also creates a new, critical teaching challenge: Design educators must help students learn how to guide generative AI tools and evaluate AI generated outcomes, not just carry out tasks.

Prompt literacy and emerging competencies

(identified by 5 participants: ID-DE 3, ID-DE 4, DK-DE 1–2, ES-DE 3)

Prompt literacy is the ability to formulate effective and critical language used to direct AI systems. Prompt literacy has emerged as a new professional competency. Danish lecturers require students to submit AI-usage statements after each project, outlining which prompts were used and why. “They must justify their prompts and document how they engaged AI to support design thinking” (DK-DE 1).

In Indonesia, design educators used similar strategies to encourage design process transparency: “If they use AI, they must explain what part AI contributed to” (ID-DE 3). Such transparent and reflective documentation helps students develop meta-cognitive awareness and a clearer understanding of how their thinking and language choices shape creative outcomes.

Participants described this as an evolving literacy that blends technical, ethical and linguistic skills. As one design educator noted: “Prompting is not just a command, it’s communication. It shows how well you think” (DK-DE 1). This emphasis signals a broad pedagogical shift: Teaching generative AI use now involves cultivating critical articulation, not simply operational mastery.

Ethics and authorship, and assessment integrity

(identified by 7 participants: ID-DE 1–4, DK-DE 1–2, ES-DE 2)

Concerns about plagiarism, authorship and ethical awareness were voiced by nearly all participants. While not every design educator raised these issues explicitly, the broader discussion reflected a shared view that ethical reflection is a pedagogical responsibility.

In Indonesia, students openly resisted generative AI use in one class, calling it “plagiarism” when peers mimicked recognisable visual styles (ID-DE 2). Design educators responded by establishing clear guidelines: “I don’t allow AI in design projects, but students may use it for research or exploration” (ID-DE 1).

In Denmark, industry demand pushes universities to integrate AI, creating a need for educators to define ethical limits: “Design agencies use it as much as possible… students must learn it, but we raise expectations for documentation and reflection” (DK-DE 2).

El Salvadoran design educators, while less exposed to industry pressure than Denmark, emphasised the moral dimension of authorship in broader terms. They stressed that design requires independent judgement and cannot be outsourced: “The student should still think; AI cannot replace that” (ES-DE 2).

All interview participants agree that ethical reflection is a pedagogical responsibility, not an optional debate. Assessment frameworks increasingly include sections on AI transparency, reinforcing the value of ethical literacy alongside creative skills.

Infrastructure, access, and institutional readiness

(identified by 7 participants: ID-DE 1–4, ES-DE 1–2, DK-DE 1)

Access to generative AI technology emerged as a material constraint in Indonesia and El Salvador. Participants cited the cost of premium AI tools and little institutional financial support: “Subscriptions are not accessible for everyone here; some lecturers pay themselves” (ID-DE 4). Another noted: “We have no policy or funding for this yet. It all depends on individual initiative” (ES-DE 1).

These inequities shape not only what tools are used but also how generative AI is positioned pedagogically. In many cases, limited access leads to more reflective, conceptual uses of AI rather than production-oriented applications. Indonesian design educators, for example, focus on low-cost or open-source platforms, integrating AI into research discussion rather than image generation.

In contrast, Danish participants noted that their institutions encouraged experimentation with AI tools and provided resources to support it, making it easier to integrate generative AI into studio activities. The contrast underscores how material conditions mediate pedagogical innovation, revealing the uneven global landscape of AI adoption in design education.

Co-learning, reflection, and educator adaptation

(identified by 5 participants: ID-DE 2–3, DK-DE 1–2, ES-DE 1)

A final theme concerns how design educators themselves are learning alongside students. Several design educators acknowledged the need for iterative experimentation. As one Indonesian design educator put it: “We are learning together… AI changes too fast to claim expertise” (ID-DE 2). Danish design educators echoed this view, describing AI pedagogy as “a co-learning process” requiring openness and flexibility (DK-DE 1). An El Salvadoran design educator framed AI skills as part of an evolving professional identity: “We are not only teaching design but learning new ways of thinking about it” (ES-DE 1).

The theme of co-learning highlights educators’ reflexive adaptation to generative AI. They are not only observers of technological change but active participants negotiating its meaning. This theme shows that generative AI integration transforms academic practice and student learning in parallel, as both design educators and students adjust their ways of thinking and making as the technology evolves.

Synthesis and discussion

Collectively, these themes frame generative AI as a force that significantly transforms design education. Participants consistently reaffirmed the importance of human judgement and reflection, while design educators in Indonesia and El Salvador additionally highlighted manual craftsmanship as central to design learning. Across all institutions, design educators agreed that AI introduces new cognitive, linguistic and ethical dimensions to creative work.

The findings reveal regional asymmetries, as resource constraints in El Salvador and Indonesia foster adaptive and reflective pedagogies, whereas stronger infrastructure in Denmark encourages integration and professional alignment. Yet across all institutions, educators agree on maintaining design as a critical, human-centred discipline, even as its methods evolve through collaboration with generative AI systems.

Reframing generative AI in design education: From technical tool to varying modes of cognitive support

The findings of this study show that generative AI occupies different cognitive and pedagogical roles across Indonesia, El Salvador and Denmark. Rather than being understood uniformly as a “cognitive partner,” educators expressed varied levels of acceptance, caution and critical engagement depending on their institutional and cultural environments.

In Denmark, participants sometimes described AI as something students can “think with,” reflecting a context where generative AI use is already embedded in industry practices and supported by strong infrastructure. This conceptual framing took shape through reflective documentation requirements: Danish design educators asked students to submit statements detailing which prompts were used and why. These statements were not merely procedural; they were intended to make students’ reasoning visible and to reinforce AI as a tool for thinking rather than a shortcut.

Some Indonesian design educators prohibited the use of generative AI in design projects or limited its use to research or preliminary exploration of ideas. When generative AI was permitted, design students were required to explain its role because design educators feared plagiarism, over-reliance on quick AI responses and the erosion of manual craft. In both contexts, documentation functioned as a metacognitive strategy, positioning generative AI as part of a reflective dialogue rather than as an unquestioned generator of ideas.

By contrast, El Salvadoran educators emphasized ethical reflection and originality but did not require structured written statements. Instead, they relied on verbal justification during critiques to ensure that generative AI use did not replace reflective thinking or diminish the discipline of making.

This nuanced framing aligns with emerging research arguing that generative AI is shifting from task automation toward more interactive forms of cognitive support (Fleischmann, 2025a; Gökdemir and Kayan, 2025; Lloyd et al., 2022). While this framing aligns closely with the Danish case, design educators in Indonesia and El Salvador adopted a stance grounded in ethical vigilance, cultural preservation and foundational studio pedagogy. Here, generative AI operates primarily as a catalyst for explanation and reflection rather than as a dialogic partner.

This variation contributes a theoretical refinement to distributed cognition perspectives (Clark and Chalmers, 1998). Although generative AI can externalise reasoning or support metacognitive insight, the extent to which it does so is deeply shaped by pedagogical traditions and resource conditions. Danish design educators used generative AI to expand conceptual exploration; Indonesian and El Salvadoran design educators used it to reinforce reflective practice and maintain control over the learning process.

Manual-first pedagogy and the persistence of tacit knowledge

Another key finding of this research highlights the role of manual-first pedagogy in design education. In Indonesia and El Salvador, manual techniques such as sketching, collage, and model-making continue to anchor cognitive development, despite AI’s increasing presence. These practices align with Polanyi’s (1966) notion of tacit knowledge as embodied, intuitive dimension of knowing that cannot be formalised. As Indonesian and El Salvadoran design educators noted, “manual work builds sensitivity” and “discipline,” grounding creative reasoning before students engage computational tools. Findings in this study reveal manual-first teaching as a form of epistemic preservation and a deliberate strategy to maintain reflection and craftsmanship amid increasing automation. Fleischmann (2025b) similarly argues that traditional studio practice cultivates empathy and intuition, qualities that risk erosion if generative AI is used uncritically.

The intersection of manual processes and AI experimentation suggests that hybrid pedagogy may shape the future of design education. This approach integrates embodied learning with algorithmic reasoning. Students in manual-first contexts cycle between physical making, reflecting, and directing AI, echoing the iterative learning patterns described in design studio theory (Kolb, 1984; Schön, 1983) and extended to AI-mediated processes in recent work on Design Thinking (Gonsalves, 2024). Rather than viewing low-tech environments as deficient, this study reframes them as critical spaces for preserving slow, reflective modes of learning. The manual-first stance becomes a pedagogical strength, anchoring design education in sensitivity, materiality, and cultural context.

Cognitive reconfiguration: Directing, prompting, and meta-learning

Participants across all contexts observed a clear shift in cognitive roles from designers as makers to designers as directors or curators of AI generated outcomes. This transformation aligns with Gökdemir and Kayan’s (2025) view of AI as a cognitive co-agent and is supported by Jiang’s (2025) argument that authorship in AI-mediated design increasingly takes the form of “curatorial negotiation” rather than manual execution.

Findings extend this theoretical insight by showing that prompt literacy functions as a metacognitive bridge between concept and computation. In both Denmark and Indonesia, educators required students to submit statements explaining their AI use and reasoning, a practice consistent with research emphasising reflective documentation in AI-supported learning (Fleischmann, 2025a).

Through these activities, prompting becomes more than a technical exercise; it develops into a linguistic and ethical discipline. As Danish design educators observed, students learn “how to think aloud” when prompting, revealing their assumptions and clarifying their decision-making. This supports the view that AI literacy involves understanding how and why outputs are produced, a capacity aligned with what Muji et al. (2023) refer to as epistemic transparency in AI-supported learning.

Consequently, this study reframes AI integration as a form of meta-learning in which students learn not only to use AI tools but also to reflect on their own thinking and creative reasoning. This reflexive process positions design education as an ongoing dialogue between human intention and algorithmic interpretation.

Ethics, authorship, and the reconfiguration of responsibility

Design educators consistently linked AI integration to new ethical challenges surrounding authorship, originality, and integrity. These concerns mirror those documented in global studies (e.g., Amini, 2025; Bartlett and Camba, 2024; Fleischmann, 2025a) but take on distinct cultural expressions. Indonesian participants equated AI plagiarism with moral failure, while Danish design educators emphasised documentation and accountability, and El Salvadoran design educators highlighted authenticity as a measure of human creativity.

These differences illustrate how ethical reasoning is contextually mediated and socially constructed. As Zeivots et al. (2025) argue, ethical deliberation must be embedded in collaborative design practice rather than treated as a stand-alone procedural requirement. Findings confirm that design educators are embedding ethical reflection directly within design critiques, prompting students to justify not only how they used AI but why.

The resulting shift toward distributed authorship, where creative ownership is negotiated among human and machine contributors, marks a major conceptual evolution. Gökdemir and Kayan (2025) and Fleischmann (2025b) both note that authorship in AI-mediated design is shared and context-dependent rather than individually defined. The empirical evidence supports this and further demonstrates that distributed authorship is already shaping design assessments. Design educators now require reflective statements that account for AI’s role, effectively integrating ethics as an assessable skill.

This pedagogical move reframes ethics as a form of epistemic vigilance, which Rana et al. (2025) describe as the ability to question the accuracy and reliability of AI generated outputs. The present study extends this perspective by showing how ethical awareness can also operate as a form of creative literacy in design education.

Inequality, infrastructure, and pedagogical adaptation

The role of infrastructure and access to generative AI technology emerged as a defining pedagogical difference. In the Indonesian and Salvadoran contexts, limited hardware, subscription costs and the absence of formal institutional policy restricted access to generative AI tools. Similar constraints are documented in broader international research, where infrastructure gaps and cost barriers hinder equitable AI adoption in less-resourced contexts (Henadirage and Gunarathne, 2024; UNESCO, 2024). Yet this study shows that constraint can also foster distinctive forms of innovation. Design educators working in resource-limited environments tended to approach AI reflectively and conceptually rather than procedurally, encouraging design students to think critically about why and how AI should be used before focusing on what it can produce.

This finding challenges assumptions embedded in Euro-American literature that technological maturity naturally leads to pedagogical advancement. Instead, it suggests that material scarcity can sharpen criticality. When tools are limited, reflection can become deeper and more deliberate. In this way, inequity can function as an unexpected catalyst for pedagogical resilience–a perspective that is largely absent from the current literature.

The adaptive strategies described by participants echo UNESCO’s call for more inclusive and context-aware approaches to AI in education, where diverse institutions help shape norms around technological use (Miao and Holmes, 2023). By foregrounding perspectives from El Salvador and Indonesia, this study expands the epistemic landscape of design education research and resists homogenising narratives of technological progress that often dominate Western discourse.

Conclusion: Summary of key implications

This study contributes to the growing discourse on generative AI in design education by highlighting perspectives that are rarely represented in the literature. Drawing on interviews with design educators from Indonesia, El Salvador and Denmark, it examines how AI is interpreted and integrated within different cultural, material and institutional conditions. Including Denmark provided a useful point of contrast with a well-resourced European setting, clarifying how infrastructure and pedagogical traditions shape educators’ responses to AI.

Findings show that integrating generative AI into design education is a layered process shaped by setting, tradition and reflective practice. Rather than replacing design’s long-standing attention to creativity and ethics, AI sharpens these concerns. Design educators across the three settings generally viewed generative AI not as a replacement for human creativity but as a tool that can support, extend or provoke students’ thinking. They emphasised reflection, authorship and ethical awareness as central to guiding its use. Prompt literacy also emerged as a foundational skill, because design students must learn to articulate their ideas clearly to guide, evaluate and justify AI-generated outcomes.

Manual-first traditions in Indonesia and El Salvador continue to anchor early design learning. Hands-on making and tacit knowledge were described as essential for cultivating sensitivity and judgement, providing a foundation from which students can make informed curatorial decisions about AI-generated material. These approaches offer alternative models of innovation that prioritise reflection over speed. In Denmark, strong digital infrastructure and alignment with industry expectations supported more structured experimentation and curricular integration. Yet Danish design educators also required documentation and justification, ensuring that AI use strengthens rather than weakens professional standards. The contrast across these three settings underscores how material conditions mediate pedagogical innovation and highlights the uneven global landscape of AI adoption in design education.

The findings further suggest that design pedagogy itself is shifting. Design educators are reconsidering their roles and methods, integrating generative AI not as an automatic solution but as a catalyst for discussion, critical questioning and ethical decision-making. This development builds on design education’s reflective traditions, which have long balanced creativity with responsibility. Generative AI amplifies this balance by prompting renewed attention to how designers think, reflect and decide alongside generative AI systems.

Throughout the study, generative AI is treated not as an inevitable or neutral force but as a technology whose adoption is actively negotiated by educators within specific ethical, cultural, and institutional contexts. Whether design studio education will ultimately shift from doing to directing remains an open question, as design educators and students explore how to balance hands-on and digital making with evaluating and interpreting the ideas generated by AI. The challenge for design educators lies in preserving reflection and tacit understanding while using generative AI to extend rather than replace human creativity.

While this study is grounded in design education, several of its implications extend to adjacent arts and humanities fields that engage in studio-based, process-oriented learning. In disciplines such as studio art, architecture, media arts, and creative writing, generative AI similarly unsettles established notions of authorship, process visibility, and assessment integrity. Practices emphasised by participants in this study, including documentation of decision-making, justification of tool use, and critical evaluation of AI-generated material, are therefore likely to be relevant beyond design education.

At the same time, discipline-specific traditions, such as material practices, performance-based authorship, or literary notions of originality, mean that AI integration will unfold differently across fields. Rather than operating as a uniform pedagogical intervention, generative AI takes on meaning through existing disciplinary values and assessment cultures. Across the arts and humanities, its use in early ideation raises shared questions about authorship claims and originality in terms of creative contribution and agency, highlighting the need for transparent processes and critical judgement when AI-generated material is incorporated into creative work.

Future research should follow these developments across longer time scales and across diverse educational regions. Further work could investigate how ethical negotiation and authorship are taught, how tacit and embodied knowledge persist or change, and how assessment frameworks adapt to AI-enabled processes Research might also explore how co-creation practices that involve human–AI collaboration influence creativity, judgment and studio dynamics. Ultimately, the challenge for design education is not simply to adopt generative AI but to preserve the human-centred values of the discipline by preparing designers who can think critically, act ethically and create reflectively in partnership with intelligent generative systems.