The design space of the basic design studio: An analysis and assessment with synthetic solutions

Abstract

The basic design studio is considered as the core of the design education in terms of teaching novice designer how to reason for design. Despite its crucial role, the first-year design students often recorded to have difficulties to understand the abstract concepts and principles conveyed in the studio by the design problems. These problems present challenges for students due to implicit definitions and abstract concepts given in the assignment briefs. Because the students have neither previous experience in dealing with abstract design problems that require elaborating multiple and interrelated aspects of the problem at once, nor developed a reasoning mechanism to generate design solutions. Thus, this research aimed to search for a method to help students better understand the abstract concepts and principles conveyed the in the basic design studio and explore the impact of the design problems on the generated solutions by a text-to-image artificial intelligence (AI) model. It is hypothesized that the concept of the design space can offer a medium for the students to reconstruct the design problems to generate solutions since the concept inherits problem and solution spaces bonded with the design process. By reviewing the literature the concept reframed in the scope of the study as: assignment briefs constituting the problem space with inherent ill and well definitions; solution space contains the design process outputs. To explore the impact of the problem space on the solution space, a series of synthetic design solution spaces were generated using a text-to-image diffusion model. The selection of the AI model was critical for the study. The text-to-image diffusion models are assumed in this study as suitable for assignment-based design education that emphasized learning-by-doing type through solution assessment and development during a generative process. Hence, the synthetic design solution spaces are created as alternative assessments to elucidate the problem definitions in a sample set of assignment briefs and to consider the impact of the brief and the feedback process on design solution space generation. The methodology encompasses a retrospective perspective in terms of using two sets of problem spaces of two different design schools to generate a series of synthetic design solution spaces. In total three solution spaces were generated for each problem space analyzed in the study. These solution spaces had substantial differences in terms of including a feedback mechanism. In the first step, the analyzed two sets of assignment briefs were translated into text prompts by preserving their semantic organization. These prompts were used to generate the first and second synthetic solution spaces that correspond to the two problem spaces of different institutions. The first solution spaces were subjected to an evaluation process by design experts, alluding to a feedback/critique session in a conventional design studio, whereas the second solution spaces were kept as a control group for the further assessment process. Secondly, text-prompts were revised based on the feedback and the third synthetic solution space was generated using the diffusion model. Lastly, the performances of the generated synthetic solution spaces, including the control groups, were evaluated in semi-structured design expert interviews. The expert interviews indicated a retrospective discussion in terms of comparing the visual impacts of the explicitness of the problem spaces on the synthetic design solution instances. Overall findings indicated that the performances of the generated solutions tend to increase when the brief defines the problems explicitly. Besides, the feedback process enhances the overall performance of the design solution spaces, as they introduce the implicit agenda of the briefs defined with the ill-defined design problems. Although the assessment results indicated several limitations of the model for representing well-defined design problems, experts evaluated the performance of the model as promising to elucidate the ill-defined problems for the students. Thus, with expert guidance, synthetic solution spaces can be used to expose students to a large number of solutions as they interpret the given design problem, the principles, and key concepts, and develop critical perspectives on their process and productions. Moreover, the potential implementation strategies of the AI tool in the first-year design studio were discussed in terms of enlarging the design space of the novice designer, to enable them to develop a reasoning mechanism.