Generative AI in Design Thinking Pedagogy: Enhancing Creativity, Critical Thinking, and Ethical Reasoning in Higher Education

Synthesis: Mixed-methods analysis of 112 student reflections from a 12-week course examining how Generative AI tools (ChatGPT, DALL·E) operate across Design Thinking's five stages. Four themes emerged: Perceived Benefits (enhanced Creativity and Accessibility), Ethical Concerns (bias and authorship ambiguity), Hesitance & Acceptance (evolution from skepticism to strategic adoption), and Critical Validation (development of epistemic vigilance). Sentiment analysis showed 86% positive responses, while ethical concerns generated significant negative sentiment (62%). The authors conclude that Generative AI, when pedagogically scaffolded, augments rather than replaces human judgment — students evolved from passive users to critical evaluators. Published in JUTLP, DOI https://doi.org/10.53761/tjse2f36.

Key Findings

• Four themes across the design process. Thematic analysis of 112 student reflections surfaced four central themes — Perceived Benefits, Ethical Concerns, Hesitance & Acceptance, and Critical Validation — each mapping to distinct dimensions of Generative AI integration within Design Thinking Pedagogies and Teaching Strategies.
• Perceived Benefits (n=94) with 86% positive sentiment. Students reported enhanced Creativity, broader ideation, and greater accessibility, positioning GenAI as a catalyst for divergent thinking during the ideation stage rather than a simple answer-provider.
• Ethical Concerns (n=66, 59%) drove 62% negative sentiment. Central issues were algorithmic bias (e.g., DALL·E reproducing gender stereotypes) and authorship/ownership ambiguity — but the negative sentiment reflected active ethical engagement, not rejection of the technology.
• Hesitance & Acceptance (n=56, 50%). Students moved from skepticism (seeing AI as "cheating") toward conditional, strategic adoption, reaching 72% positive orientation by course end — a shift that required instructor-led Scaffolding and peer dialogue.
• Critical Validation. Students developed epistemic vigilance, learning to verify outputs, detect bias, and validate sources — evolving from passive users into critical evaluators of AI-generated content.
• Augment, don't replace. When pedagogically scaffolded within Design Thinking, Generative AI augments rather than replaces human judgment, supporting Creativity while preserving learner Learner Agency.

Study Design & Method

The study was conducted across a 12-week undergraduate Design Thinking course at a metropolitan Australian university, with 112 students drawn from business (60%), education (19%), law (12%), and interdisciplinary programs (9%). None had prior formal training in Design Thinking or Generative AI, providing a consistent baseline. GenAI tools were embedded at each of the five design stages: ChatGPT for empathise (persona Simulation), AI summarization for define, DALL·E and ChatGPT for ideate, DALL·E for prototype, and AI analytics for test.

The study adopted a mixed-methods paradigm guided by a Constructivism epistemology grounded in Experiential Learning, Vygotsky's sociocultural theory, and Dewey's pragmatism. Qualitative analysis used Braun and Clarke's six-phase thematic analysis framework with NVivo 14, achieving high intercoder reliability (Cohen's κ = 0.82 on a 20% subsample). Sentiment analysis was implemented via a fine-tuned DistilBERT model (validated at 89% accuracy, F1 = 0.85), cross-checked with manual coding (88% concordance) and the VADER lexicon.

What this means for practice

• Instructors. Integrate GenAI deliberately at each stage of the Design Thinking process with critical framing and structured team-based work; acceptance in this study came from intentional guidance, not exposure alone.
• Instructors. Build output interrogation into the task — verifying claims, detecting bias, and tracing sources — and run explicit activities on cases such as DALL·E reproducing gender stereotypes rather than leaving Ethics implicit.
• Faculty developers. Treat AI Literacy as a multidimensional competence spanning Creativity, Ethics, and critical reasoning rather than technical proficiency, and support staff to design GenAI-integrated Higher Education environments that promote learner Learner Agency instead of automation.
• Designers. Position GenAI as a co-creator within Human AI Collaboration and use scaffolds that keep human judgment at the center, countering Cognitive Offloading and automation bias.

Limitations

• The four themes were derived primarily from 112 student reflections, and the authors state that self-report methods carry social desirability, retrospective-reconstruction, and weak-correlation-with-learning risks.
• The study ran in a single 12-week undergraduate design course at one metropolitan Australian university, so institutional variables such as faculty attitudes, digital infrastructure, and curricular autonomy may limit generalizability.
• No objective outcome measure was used, so the reported gains in Creativity and ethical reasoning remain student claims; the authors note the absence of pre/post achievement or cognitive-load testing.
• Sentiment analysis (a fine-tuned DistilBERT validated at 89% accuracy) surfaced affective trends, but the study did not examine how emotions such as anxiety or confusion shape engagement with GenAI.

Citation

Rana, V., Verhoeven, B., & Sharma, M. (2025). Generative AI in design thinking pedagogy: Enhancing creativity, critical thinking, and ethical reasoning in higher education. Journal of University Teaching and Learning Practice, 22(4).