Date of Defense
1-6-2026 10:30 AM
Location
Room 1124, Building F1
Document Type
Dissertation Defense
Degree Name
Doctor of Philosophy in Architectural Engineering
College
College of Engineering
Department
Architectural Engineering
First Advisor
Ahmed Agiel
Keywords
Artificial Intelligence, Architectural Education, Repertory Grid Technique, Design Autonomy, MidJourney AI, Pedagogical Innovation, Brand Image, decoding preferences
Abstract
Architectural education is currently at a crossroads. On one hand, technological innovation, particularly in the realm of design tools, is advancing at a pace that few curricula can match. On the other, public expectations of architecture are shifting, with communities seeking designs that better reflect their identity and lived experience. Yet, in many universities, the traditional studio model still dominates, relying on instructor-led critique sessions that, while valuable, can sometimes narrow student independence and overlook community priorities. This dissertation takes up that challenge by exploring an unusual pairing: generative AI text-to-image platforms, such as MidJourney, with the Repertory Grid Technique (RGT), a method rooted in psychology for eliciting nuanced design constructs. The goal is not merely to test a new tool but to see whether this integration can help students exercise greater autonomy while producing brand image designs that genuinely resonate with local communities. The study unfolded in three phases. First, RGT interviews were conducted with 629 residents (257 from Al Ain) resulting in a rich set of constructs that capture the community's visual language, aesthetic preferences, and cultural markers. These were not just tallied; they were processed using natural language processing, hierarchical clustering, contradiction mapping, and qualitative refinement to decode the community design preferences in a single list, ensuring the list was both precise and representative. In the next phase, students from three UAE universities(UAEU, Amity Dubai, and AURAK) applied their own RGT elicited constructs in combination with Al Ain local image reference photos using Midjourney in real design studios. This represents a teaching framework by generating design scenarios from AI which inspires the students to design for Al Ain brand image. At UAEU, 19 students worked individually with MidJourney, later UAEU students reflected on their process in structured interviews for developing the framework. Later on, the developed framework was applied at Amity Dubai (n=11) and AURAK (n=12). Autonomy was assessed through survey and a tailored version of the Autonomous Learner Scale among the three universities, then it was compared with students who didn't participate. While brand image achievement was assessed through consulting a panel of architects (n=12) from the UAE to assess the students design based on the decoded list of community design preferences from phase 1, and compare them with designs for students who didn't participate. What emerged was interesting; the Welch's t-test for ALS shows a significant improvement in students' autonomy for the group that participated in the framework. More importantly, the resulting projects showed a closer alignment with the community's aesthetic. This work contributes a decoded preference list for Al Ain community and a large dataset for future studies (629 RGT interviews with UAE citizens about their architectural preferences). Additionally. it offers a new framework that merges AI text-to-image tools with psychologically grounded design methods, offering educators practical ways to encourage independent thinking while maintaining cultural responsiveness. Its novelty lies in blending generative AI, educational psychology, and architectural pedagogy into a systematic teaching strategy. Demonstrating that AI, when guided by RGT insights, can strengthen rather than dilute the human dimension of architectural education.
Included in
AI and Repertory Grid in Architectural Studios for Crafting a Brand Design
Room 1124, Building F1
Architectural education is currently at a crossroads. On one hand, technological innovation, particularly in the realm of design tools, is advancing at a pace that few curricula can match. On the other, public expectations of architecture are shifting, with communities seeking designs that better reflect their identity and lived experience. Yet, in many universities, the traditional studio model still dominates, relying on instructor-led critique sessions that, while valuable, can sometimes narrow student independence and overlook community priorities. This dissertation takes up that challenge by exploring an unusual pairing: generative AI text-to-image platforms, such as MidJourney, with the Repertory Grid Technique (RGT), a method rooted in psychology for eliciting nuanced design constructs. The goal is not merely to test a new tool but to see whether this integration can help students exercise greater autonomy while producing brand image designs that genuinely resonate with local communities. The study unfolded in three phases. First, RGT interviews were conducted with 629 residents (257 from Al Ain) resulting in a rich set of constructs that capture the community's visual language, aesthetic preferences, and cultural markers. These were not just tallied; they were processed using natural language processing, hierarchical clustering, contradiction mapping, and qualitative refinement to decode the community design preferences in a single list, ensuring the list was both precise and representative. In the next phase, students from three UAE universities(UAEU, Amity Dubai, and AURAK) applied their own RGT elicited constructs in combination with Al Ain local image reference photos using Midjourney in real design studios. This represents a teaching framework by generating design scenarios from AI which inspires the students to design for Al Ain brand image. At UAEU, 19 students worked individually with MidJourney, later UAEU students reflected on their process in structured interviews for developing the framework. Later on, the developed framework was applied at Amity Dubai (n=11) and AURAK (n=12). Autonomy was assessed through survey and a tailored version of the Autonomous Learner Scale among the three universities, then it was compared with students who didn't participate. While brand image achievement was assessed through consulting a panel of architects (n=12) from the UAE to assess the students design based on the decoded list of community design preferences from phase 1, and compare them with designs for students who didn't participate. What emerged was interesting; the Welch's t-test for ALS shows a significant improvement in students' autonomy for the group that participated in the framework. More importantly, the resulting projects showed a closer alignment with the community's aesthetic. This work contributes a decoded preference list for Al Ain community and a large dataset for future studies (629 RGT interviews with UAE citizens about their architectural preferences). Additionally. it offers a new framework that merges AI text-to-image tools with psychologically grounded design methods, offering educators practical ways to encourage independent thinking while maintaining cultural responsiveness. Its novelty lies in blending generative AI, educational psychology, and architectural pedagogy into a systematic teaching strategy. Demonstrating that AI, when guided by RGT insights, can strengthen rather than dilute the human dimension of architectural education.