How to Support Situated Design Education through AI-Based Analytics
Abstract
Design education is essential to pedagogy across fields. It helps develop creative skills and abilities. To meet growing demand for design education, researchers are investigating novel means for assessing and providing feedback on students’ design projects. To complement instructors’ work, we investigate the potential of artificial intelligence (AI), which has matched humans in performing complex tasks and activities.
We build on prior research by Suchman and Dourish in showing how developing useful AI support requires understanding situated practice. We engage instructors in co-design. In theorizing, we twice invoke creative cognition’s family resemblance principle, first, to contribute new understandings of uses and limitations of design rubrics, and then to identify an analogous role for AI-based design creativity analytics: assessing no particular characteristic is necessary or sufficient; each only tends to indicate good design work. We contribute situating analytics, a paradigm for conveying the meaning of measures that align with design rubrics, by contextually integrating the presentation of measures with associated design work.
We develop results across fields. We integrate the measurement and presentation of multiscale design characteristics that provide insights into students’ use of space and scale. We find that situating analytics supports instructors in understanding what the measures mean. Through quantitative analysis, we establish the baseline performance for AI recognition of multiscale design characteristics. Through qualitative analysis of instructors’ experiences in situated course contexts, we derive implications for conveying the meaning of measures.
Subject
design educationsituated practice
co-design
multiscale design
artificial intelligence
spatial clustering
learning analytics dashboard
Citation
Jain, Ajit (2021). How to Support Situated Design Education through AI-Based Analytics. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /195556.