ARCHIVES
VOL. 7, ISSUE 2 (2025)
Evolving methodological trends in generative AI research for design and education: A Systematic review and future directions
Authors
Xu Yang, Lay Kee Ch ng
Abstract
Generative AI (GenAI) is rapidly reshaping
studio- and project-based learning within design disciplines, yet its
methodological foundations remain uneven. This systematic review examines how
research methods have evolved at the intersection of GenAI, design, and
education. Following PRISMA 2020 procedures, major databases were searched for
studies published between 2020 and 2025. From 100 records identified, 61
peer-reviewed studies were retained for double-coded analysis across research
design, instruments, analytical techniques, tool–task alignment, educational
contexts, and markers of rigor. The corpus is dominated by secondary syntheses,
with systematic and bibliometric reviews comprising 41.0% of studies, while
empirical research remains less common (quantitative: 19.7%; qualitative:
13.1%; mixed methods: 4.9%). A sharp acceleration in publications has occurred
since 2023. Although 16 studies reported efficiency gains—such as reductions of
approximately 40–55% in visualization turnaround time and 1.5–2 times more
design iterations—the generalizability of these findings is constrained by
reliance on small-scale, single-site pilots. Methodological transparency and
rigor are recurring weaknesses: 68.9% of studies describe analytical procedures
only in generic terms, 39.3% omit survey or observation instruments, 60.7% fail
to specify AI tool versions or prompting strategies, and virtually none report
preregistration or replication. Inter-coder reliability was documented in only
3.3% of cases. In addition, outcome measures remain heterogeneous and rarely
standardized across process-level indicators (e.g., prompt–response
trajectories, density of peer critique) and product-level metrics (e.g.,
expert-rated creativity with reliability checks).On this basis, the review
proposes a forward-looking agenda: (i) enforce transparent reporting using
standards such as PRISMA and TOP; (ii) expand the use of quasi-experimental and
mixed-methods designs with longitudinal, multi-institutional samples; (iii)
establish standardized measurement frameworks that integrate process and
product indicators; (iv) align tools with tasks, for instance applying language
models to critique and text-to-image systems to ideation, supported by explicit
scaffolds; and (v) embed ethics-by-design and open science practices, including
prompt logs, datasets, and rubrics, to enable reproducible and equitable
scholarship. Taken together, this roadmap positions GenAI not merely as a
technical accelerator but as a methodologically robust and socially responsible
catalyst for advancing design pedagogy.
Download
Pages:91-100
How to cite this article:
Xu Yang, Lay Kee Ch ng "Evolving methodological trends in generative AI research for design and education: A Systematic review and future directions". International Journal of Social Research and Development, Vol 7, Issue 2, 2025, Pages 91-100
Download Author Certificate
Please enter the email address corresponding to this article submission to download your certificate.