Human–AI Collaborative Design in Architectural Studios: Evaluating Paradigm Shifts Across the Six Stages of the Design Process

Abstract

Artificial intelligence (AI) is increasingly transforming architectural education, shifting design studios toward human–AI collaborative workflows. This study investigates the impact of AI integration across the six stages of the architectural design process: pre-design, conceptual design, schematic design, design development, documentation, and presentation. A mixed-methods approach was adopted, combining survey data from 17 master’s degree students with reflective insights from eight faculty members involved in hybrid AI-supported studio environments. AI’s influence was evaluated using six indicators: efficiency, creativity enhancement, accuracy, interdisciplinary integration, adoptability, and environmental or architectural impact. The findings indicate that AI is most effective during early design stages, where it supports idea generation, visualization, and rapid iteration. Its impact becomes less pronounced in later technical phases, where human expertise and critical reasoning remain essential. Students perceived AI as a creative catalyst and productivity enhancer, while faculty emphasized its analytical and evaluative potential in supporting informed decision-making. Overall, AI functions most effectively as a complementary partner rather than a replacement for human agency. The study proposes a structured framework to guide ethical and pedagogically sound AI integration within architectural design studios.

1. Introduction

Architectural design studios represent the most critical educational environment in which future architects develop their design competencies, critical thinking abilities, and professional methodologies [1]. As the foundation of architectural education, the studio operates as an experimental platform where students engage in iterative problem-solving, spatial reasoning, and creative exploration within a practice-based learning framework. Beyond skill acquisition, design studios shape methodological thinking and professional identity, making them central to the transformation of architectural pedagogy.

The rapid advancement of artificial intelligence (AI), particularly generative AI (GAI) systems, is introducing a profound shift in how design knowledge is produced, evaluated, and communicated. AI technologies now extend beyond computational drafting assistance to influence ideation, visualization, simulation, and performance-based analysis. Recent research has explored hybrid AI frameworks that integrate analytical and numerical modeling with data-driven techniques to enhance design reliability and performance evaluation. These approaches indicate a shift toward more comprehensive AI integration beyond isolated design assistance tools [2].

Consequently, the architectural studio is increasingly becoming a site of human–AI collaboration rather than purely human-driven design activity. This evolution necessitates a deeper understanding of how AI reshapes learning dynamics, cognitive processes, and design workflows within higher education institutions [3]. Scholars argue that AI can enhance learning and creativity by generating alternative solutions, supporting iterative exploration, and providing immediate feedback that informs decision-making [4].

Nevertheless, the integration of AI into design education remains contested. Concerns persist regarding the potential erosion of students’ creative authorship and critical reasoning if excessive reliance on algorithmic outputs diminishes reflective engagement [5]. Moreover, the widespread accessibility of generative AI tools challenges traditional assessment practices and raises questions about academic integrity and evaluation standards [6]. These tensions reflect a broader uncertainty about whether AI acts as an empowering cognitive partner or a disruptive force within design pedagogy.

Recent contributions from the 2025 edition of the JIDA Conference (Jornadas sobre Innovación Docente en Arquitectura) provide significant insights into the pedagogical integration of artificial intelligence within architectural education. For instance, a study by Nicoleta Alexandra Simina explored the role of AI within a Research by Design (RbD) framework, examining its application across multiple stages of the design process, including conceptual exploration, visual iteration, and reflective dialogue. The study adopted a practice-based pedagogical approach, emphasizing learning through action and critical reflection. The findings indicate that AI can function not merely as a generative tool but as a “critical interlocutor” that enhances reflective thinking and supports iterative design development. However, the study also highlights key challenges, particularly the ethical implications of AI use and the risk of diminishing students’ independent judgment if not properly guided. These recent JIDA contributions demonstrate a clear shift in architectural pedagogy from viewing AI as a technical aid toward understanding it as an integrated component of the learning process. Nevertheless, most studies remain focused on qualitative and exploratory approaches, emphasizing student experience and reflective practice rather than providing structured, stage-based quantitative evaluation frameworks [7].

Similarly, a study presented at the International Conference of Contemporary Affairs in Architecture and Urbanism investigated the role of AI-assisted tools in architectural design studios within a university setting in Turkey. The research involved a sample of 25 architecture students and applied a comparative methodology between traditional design workflows and AI-supported processes. AI tools were integrated primarily during the conceptual and visualization stages, enabling rapid idea generation and enhanced visual representation. The results indicated a significant improvement in student productivity and engagement, particularly in early design phases, while limitations were observed in later technical stages requiring precision and detailed documentation. Additionally, the study emphasized that excessive reliance on AI could reduce critical reflection if not properly guided by instructors [8].

Despite the growing body of research addressing AI in architecture, most studies tend to focus on isolated tools, specific design phases, or technical capabilities. Limited attention has been given to examining AI integration comprehensively across the entire architectural design process within studio-based learning environments. This gap creates uncertainty regarding how AI influences different stages of design development, how its impact varies across conceptual and technical phases, and how educators can structure meaningful and ethically grounded human–AI collaboration [9].

The central research problem addressed in this study lies in the lack of a comprehensive and stage-based understanding of how artificial intelligence influences the full architectural design process within studio education. While AI tools are increasingly adopted, there is insufficient empirical evidence clarifying their differential impact on creativity, efficiency, decision-making, and critical thinking across the sequential stages of design. Without such understanding, educational institutions risk implementing AI in fragmented or unstructured ways that may either underutilize its potential or inadvertently weaken core design competencies.

Accordingly, this research aims to investigate the impact of artificial intelligence integration across all six principal stages of the architectural design process—pre-design, conceptual design, schematic design, design development, design documentation, presentation, and final critique. By adopting a mixed-method approach, the study evaluates AI’s influence through six key indicators: efficiency, creativity enhancement, accuracy, interdisciplinary integration, adoptability, and environmental or architectural impact. The study ultimately seeks to provide a structured framework for human–AI collaboration in architectural studios, supporting intelligent, ethical, and pedagogically sound integration strategies that prepare students for evolving AI-driven professional practice.

2. Literature Review

2.1. AI as a Pedagogical Tool in Architectural Design Studios

Several studies have examined the role of AI as a pedagogical tool within architectural education. Researchers suggest that AI can support learning by fostering iterative design processes and enabling adaptive learning environments. For instance, Beyan et al. proposed that AI facilitates iterative learning by offering multiple design alternatives and immediate feedback [4]. Similarly, El Moussaoui and Krois emphasized AI’s potential to create adaptive learning environments that shift the educator’s role from direct instruction toward guiding reflective learning [10]. A recent case study conducted during the 2024–2025 academic year at Zhejiang University integrated AI into a design studio through a 20 h technical module. Although students demonstrated varying levels of engagement, the study highlighted the importance of developing scalable models for AI-integrated design education [11].

2.2. Approaches to AI Integration Across the Stages of the Design Process

Research on AI integration in architectural design studios often focuses on specific stages of the design process, including pre-design, conceptual design, design development, and presentation. This stage-based approach allows for a clearer understanding of how AI supports learning and performance at different points within the architectural workflow.

Beginning with pre-design stage which includes programming activity, AI tools can support students in evaluating the relevance of their design intentions and constructing well-founded arguments to justify their decisions. This contributes to enhancing critical thinking, as students are required to critically assess AI-generated information rather than passively adopting it. A recent study developed hybrid teaching model for a 9-week course involving 24 international students. The course integrated AI technologies into architectural programming to evaluate the impact. The evaluation focused on the effectiveness of AI-assisted teaching which indicated that students developed a deeper understanding of the design process and significantly improved their analytical skills [12].

The conceptual design is the second stage of the design process which received significant attention from researchers, may be due to the abilities of AI tools in generation. Many studies explored how artificial intelligence can support creativity and idea development during this early stage. For example, a study by Sadek and Mohamed argued that artificial intelligence can be used as a conceptual design tool. Conducted in Fall 2022, it involved 34 architecture students utilizing AI tools to generate inspirational images from narratives which enhanced concept development. Their designs were compared to another group of 50 students using traditional methods. The findings indicated that the designs of AI-assisted students were more diverse and innovative concepts, highlighting AI’s potential [13]. In another study, Özorhon et al. depended on the use of AI diffusion models (e.g., Midjourney). It demonstrated how AI tools can enrich creativity and encourage critical design thinking among students [14]. Interior design also garnered attention in this stage, where a study by Karadağ and Ozarm examined the use of AI in interior design education, focusing on its role in enhancing conceptual design through tools like Midjourney. Findings showed that AI had positive impact on creative possibilities and idea development. While students recognize its value, they also stress about the need to address ethical concerns and preserve design authenticity [15].

From another perspective, many studies integrated artificial intelligence in design studios from an environmental perspective, which is regarded as a significant aspect of design development stage. Many studies examined how AI can support sustainable design by analyzing energy use, optimizing building performance, and suggesting environmentally friendly design solutions. This approach highlights the potential of AI to enhance both the educational experience and the environmental responsibility of future architects. A study by Dwijendra et al. examined the integration of AI into architectural education in Bali, focusing on sustainable practices while preserving traditional values. It highlighted how AI tools can enhance sustainable design education and address the challenges of balancing technological advancement with cultural heritage [16]. In Serbia and Montenegro, a study investigated how AI influences final-year architecture students, particularly regarding sustainability. Using surveys and comparative analysis, it shows that students’ knowledge about sustainability strongly affects their use of AI tools. The findings provide guidance for refining curricula to better integrate AI and sustainable design in architectural education [17].

However, some researchers focused specifically on the final stage, which includes rendering as essential activity. They studied the ability of AI to improve the quality, speed, and efficiency of rendering processes. For example, Ahmed and Alymani presented a pedagogical approach which incorporated AI generative tools (e.g., AI rendering). This included exploring AI tools that help to automate visualization tasks, making it easier for students and professionals to produce high-quality images and presentations. The results showed increased student engagement in addition to enhancement of learning outcomes [18].

Through a comprehensive review of previous studies, this research aims to provide an inclusive perspective on the overall impact of artificial intelligence (AI) on the entire architectural design process. Rather than focusing on a single stage, it seeks to cover all the six main stages) pre-design, conceptual design, schematic design, design development, design documentation and presentation and final critique (that students typically go through during their design projects.

To achieve the aims of this study, a set of specific research objectives was established to systematically evaluate the educational impact of integrating artificial intelligence (AI) within the architectural design studio. First, the study seeks to explore the influence of AI on each stage of the architectural design process, capturing the perceptions of both students and faculty members to gain a comprehensive understanding of its role. Second, it aims to identify how students and instructors perceive the effectiveness and usefulness of AI tools in organizing, supporting, and facilitating the design workflow. Third, the research examines which stages of the design process are perceived as being most or least effectively enhanced by AI, considering levels of engagement, satisfaction, and consensus among participants. Finally, the study investigates potential challenges and limitations associated with AI-assisted design learning, addressing both pedagogical considerations and practical constraints within the studio environment. Collectively, these objectives provide a structured framework to assess the integration of AI in architectural education and its impact on the learning and creative process.

3. Methodological Framework

The research methodology focused on evaluating AI’s impact within a hybrid architectural design studio that combined physical and digital learning environments. It evaluates AI impact through perceptual and educational performance indicators rather than measurable built outcomes. Data were collected from students and instructors through questionnaires, project evaluations, and observations. This approach enabled comparison between literature-based expectations and actual student performance. Data included both quantitative and qualitative measures to capture comprehensive insights into learning outcomes and design quality across sequential studio phases.

Analytical Matrix as an Evaluation Tool

To systematically evaluate the impact of AI, an analytical matrix was developed that links the six stages of the architectural design process (pre-design, conceptual design, schematic design, design development, design documentation, and presentation) with six evaluation indicators: efficiency, creativity, accuracy, AI integration, adoptability, and environmental/architectural impact (see

Table 1. Two-dimensional evaluation matrix (source: Authors).

		Measurement Indicators
		Efficiency	Creativity	Accuracy	AI Integration	Adoptability	Environmental/Architectural Impact
Stages of Design Process	Pre-design	AI tools boost efficiency by speeding up site analysis, program assessment, and data processing [19,20]	AI supports initial idea generation, providing diverse scenarios, though human analytical input remains essential [19,21]	Data accuracy relies on input quality and requires human verification for reliability [22,23]	AI supports early-stage decisions through data analysis and predictions [19,24]	Analytical and research tools can be easily incorporated into pre-design curricula [25]	AI helps assess environmental and contextual factors without directly shaping the design at this stage [26,27]
	Conceptual Design	AI speeds up the creation of design alternatives, improving decision-making in the conceptual phase [19,21]	Generative AI enhances creativity by offering diverse and novel design solutions during conceptual exploration [28,29]	Conceptual outputs need human refinement to meet project goals and constraints [19,21]	Generative design involves high AI engagement, shaping initial forms and spaces [19,23]	Adopting AI tools requires some training, but integration into studios is practical and achievable [24]	AI supports early sustainable design by guiding initial environmental performance considerations [26,30]
	Schematic Design	AI simulations help evaluate design alternatives, enhancing efficiency in spatial and functional assessments [19,27]	Creativity is partly directed toward functional performance rather than purely conceptual innovation [28,31]	Integrating spatial and energy analysis improves the accuracy of schematic layouts and early performance predictions [21,32]	At this stage, AI focuses more on analytical evaluation than on generating forms [32,33]	Effective use of simulation and evaluation tools requires technical proficiency [24]	AI supports environmental performance evaluation, guiding designs toward better sustainability outcomes [26,34]
	Design Development	Efficiency is maximized through the integration of building systems, detailing, and optimized workflows [21,23]	Creativity is focused on material and system innovation rather than overall form [22,29]	AI tools enhance accuracy in documenting and coordinating complex design elements [23,35]	AI supports analytical evaluations and detail optimization [35,36]	These tools are easily adopted in both academic and professional design development [23,25]	AI-supported optimization can improve environmental and material efficiency, reducing waste [26,37]
	Design Documentation	Automating drawings, schedules, and specifications greatly enhances documentation efficiency [19,23]	This stage is mainly execution-focused, with limited room for creative exploration [22,35]	AI enhances the accuracy of construction documentation, reducing human error [23,38]	AI is mainly used for automated detailing and schedule generation [12,27]	Adoption is high due to practical benefits and ease of integrating AI into documentation workflows [23,24]	Enhanced documentation quality indirectly supports environmental performance by minimizing material waste and construction errors [36,37]
	Presentation and Final Critique	AI speeds up the creation of visual presentations, 3D renderings, and interactive models, improving efficiency in communicating design intent [22,38]	AI enhances visual communication, enabling richer and more expressive architectural storytelling [28,39]	The accuracy of representations depends on input quality and requires human verification [22,38]	AI plays a significant role in rendering, AR/VR, and generative visualization [22,39]	These tools are easily adopted, allowing students to use AI for more impactful presentations [39]	Improves understanding of environmental and architectural impacts through clear visual representation [26,40]

Strong Relation  Moderate Relation  Weak Relation.

). The initial two-dimensional matrix, grounded in previous studies, served as a theoretical framework to assess performance at each stage. This matrix was then refined into a three-dimensional model using empirical findings from the studio, connecting the evaluation tools directly to the observed outcomes.

The six indicators were selected to capture pedagogical and professional considerations. Efficiency evaluates workflow, time management, and productivity; creativity reflects innovative and diverse design thinking; accuracy measures technical reliability. AI integration examines the depth of AI use as a tool, collaborator, or assistant; adoptability considers usability and ease of implementation; and environmental/architectural impact reflects AI’s role in sustainable and context-responsive design. Together, they provide a structured framework for assessing AI’s dynamic influence on learning outcomes and design quality throughout the studio process.

The matrix was developed based on previous studies, which indicate that AI’s impact varies by stage. In pre-design, AI strongly enhances efficiency but moderately affects creativity and accuracy. During conceptual design, AI’s influence peaks, supporting both efficiency and creativity. In schematic design, AI emphasizes performance analysis and accuracy, while creativity is somewhat constrained by functional priorities. Design development shows high impact on efficiency, accuracy, and adoptability. In contrast, design documentation demonstrates limited creative engagement but high technical accuracy. Finally, in presentation and final critique, AI strongly enhances visualization, rendering, and communication. Overall, AI evolves from analytical support in early stages, through creative facilitation in the middle, to technical precision in later stages, forming a dynamic and integrated influence in architectural studios.

4. Case Study Application

4.1. Sample

This research investigates the impact of artificial intelligence (AI) on the architectural design process across its various stages through a quasi-experimental approach. The study involved a purposive sample of 17 master’s degree students and eight faculty members from the Department of Architectural Engineering at Kafrelsheikh University, Egypt.

The study was conducted during the spring term of 2025 as part of the course titled Integrated Architectural Design Studio, which provided a suitable academic environment for applying AI tools within real design tasks and reflective discussions. Master’s degree students were specifically chosen over undergraduates due to their advanced academic experience and their ability to critically evaluate design-related issues, including the potential impact of AI integration. This selection ensures the reliability and depth of the feedback collected, as these students have completed foundational design courses and participated in multiple design studios, equipping them to assess AI tools’ integration with informed awareness and analytical insight.

4.2. Method

The study employed previously completed academic design projects that participants had developed during their architectural studies without the use of AI. To ensure the relevance and comparability of the data, participants were required to select projects completed within the last three academic years, typically corresponding to advanced stages of architectural education, where students possess a higher level of design maturity and technical competence. Each participant selected a prior project—such as a cultural center or exhibition complex that served as the basis for a renewed design process incorporating AI tools.

This approach enabled a direct comparison between traditional design methods and AI-assisted workflows across the same six design stages. By repeating equivalent design tasks and activities, the study was able to systematically identify the benefits and challenges introduced by AI integration. Throughout the course, participants were closely monitored, and their progress was documented across all stages, providing a structured framework for evaluating AI’s impact on the design process. To ensure consistency and reliability, the study defined specific activities associated with each stage, along with the expected outputs, forming a reference framework for evaluation. As illustrated in

Table 2. Stages of design process and its related activities (source: Authors).

Stages of Design Process	Activities
Pre-Design (Programming) Focus: Understanding the project requirements and site constraints	Research and Analysis: Collecting information about the site, context, client needs, and regulations.
	Site Analysis: Evaluating site conditions, including climate, topography, and existing structures.
	Program Development and Space Requirements: A documented architectural program that guides the conceptual design stage.
Conceptual Design Focus: Refining design concept	Brainstorming: Developing a wide range of design ideas and possibilities.
	Sketching and Diagramming: Creating rough sketches and diagrams to explore spatial relationships and forms.
	Initial Models: Building simple physical or digital models to test ideas.
Schematic Design (SD) Focus: developing the chosen design concept	Space Planning: Developing preliminary floor plans and layouts.
	Massing Studies: Exploring the building’s form and its impact on the site.
	Conceptual Renderings: Creating visual representations of the design.
Design Development (DD) Focus: Adding detail to the design and integrating technical aspects	Detailed Drawings: Developing more detailed floor plans, elevations, and sections.
	Refinement of Spatial Layout and Form: Developing the approved schematic design into more detailed and accurate spatial configurations
	Material Selection: Choosing materials and finishes for the project.
Design Documentation (DC) Focus: Preparing detailed drawings and specifications	Drawings Documentation: Producing detailed architectural drawings, including plans, sections, elevations, and details.
	Integrating Environmental and Structural Aspects: Ensuring that the design is both sustainable and structurally efficient.
	Specifications: Writing detailed descriptions of materials, workmanship, and installation methods.
Presentation and Final Critique Focus: Finalizing the design and presenting it	Final Models and Renderings: Creating detailed models and high-quality renderings to convey the design.
	Presentation Boards: Assembling visual materials that effectively communicate the design.
	Final Critique: Receiving critical input from faculty, peers, and external reviewers.

, three essential activities were identified for each of the six design stages.

4.3. Data Collection and Analysis

To ensure a coherent and reliable assessment of AI’s educational impact, the six evaluation indicators were systematically aligned with three main measurement tools: student questionnaires, project reviews, and studio observations. This structure was intended to establish a clear and meaningful connection between each indicator and its most relevant source of evidence as shown in

Table 3. Primary measurement tools for every indicator (source: Authors).

Evaluation Indicator	Primary Measurement Tools	Rationale
Efficiency	Student Questionnaire	Efficiency was evaluated based on improvements in time management, workflow speed, and overall productivity resulting from the use of AI tools. The assessment considered students’ ability to complete design tasks more effectively across different project stages.
Creativity	Project Review	Creativity was evaluated through a structured project review using a standardized rubric to ensure reliability. The assessment focused on the originality of design concepts, the diversity of approaches, and the level of innovation demonstrated in spatial organization, form, and overall design solutions. Each project was scored by multiple reviewers independently evaluated the work to enhance inter-rater reliability and reduce subjectivity.
Accuracy	Structured Studio Observation	Accuracy was evaluated through faculty observations focusing on the technical precision, consistency, and reliability of AI-assisted design outputs. The assessment examined the alignment between design intent and final representations, as well as the reduction in technical errors across different project stages.
AI integration	Structured Studio Observation	The assessment considered the breadth and depth of AI use across different design stages, the level of critical engagement with AI-generated outputs, and the degree to which AI tools were integrated into the overall design workflow.
Adoptability	Student Questionnaire	Adoptability was evaluated through students’ feedback on usability, accessibility, and their willingness to continue using AI tools in future design projects. The assessment reflected overall acceptance of AI within the studio environment.
Environmental and architectural impact	Project Review	This indicator was assessed through a project-based analytical review, combining qualitative judgment focusing on how AI-supported designs responded to environmental and contextual factors.

, and these indicators were measured using a five-point Likert scale (1 = strongly disagree, 5 = strongly agree). Efficiency and adoptability were evaluated through student questionnaires, as these aspects depend on students’ personal experiences, perceptions of productivity, and willingness to engage with AI tools.

Creativity and environmental or architectural impact were assessed through project reviews, enabling instructors to judge the originality and contextual quality of the design outcomes. Accuracy was measured using studio observations, as technical precision and adherence to professional standards require expert input. Also, AI integration was analyzed through studio observations, providing real-time insights into how effectively AI tools were embedded within each stage of the design process. This structured mapping created a balanced methodology that combined subjective and objective perspectives, integrating students’ experiential feedback with faculty evaluations and observational evidence to offer a comprehensive understanding of AI’s role in the architectural design studio.

4.4. Strategy and AI Tools Used in the Study

This study adopted a hybrid, design-studio-based methodology for 14 weeks, during which the design process stages were distributed accordingly. As shown in

Table 4. Workflow of the design studio in conjunction with the AI tools used (source: Authors).

	Week Number	Traditional Tools	Digital Tools	Out Put	AI Tools Used
	Week 1	Introduction Lecture
Pre-design	Week 2	Site analysis Manual sketches	Google Earth GIS; AutoCAD 2024 (Site Plans) Revit (Site Context)	Site analysis Data collection Problem identification	Chat GPT (data collection & analysis) Microsoft Designer (concept images) AI GIS tools (environmental analysis)
Conceptual Design	Week 3, 4	Hand sketches Physical models	Sketch Up Rhino Grasshopper	Initial design concept Sketches and diagrams	Microsoft Designer (quick ideation) Canva AI Ideogram AI Playground AI (concept images)
Schematic Design	Week 5, 6, 7	Manual plans Group discussions	AutoCAD Revit	Developed design concept, zoning, initial plans	Planner 5D AI (space layouts) Room GPT (layout suggestions)
Design Development	Week 8, 9, 10	Manual detailing Area schedules	AutoCAD Revit	Detailed architectural drawings	Autodesk Forma AI (solar, wind & environmental analysis) Krea.ai (design alternatives—basic use)
Design Documentation	Week 11, 12	Manual construction docs Material schedules	Revit (Construction Docs) AutoCAD (Details)	Architectural drawings supported by structural system	Basic AI floor plan checkers (validation only) Microsoft Copilot 2024 (text & documentation support)
Presentation and Final Critique	Week 13, 14	Printed boards PowerPoint	Photoshop Illustrator InDesign	Final project presentation	Canva AI (presentation boards) Gamma.app (presentation generation) Runway ML (simple video) ChatGPT 2024 (project narrative & explanation)

, the six design stages were arranged sequentially with the AI tools used, allowing for accurate monitoring and evaluation throughout the course period. Throughout the course, AI-driven platforms were introduced alongside digital and traditional tools. The purpose of this approach was to provide students with advanced resources that can enhance their design ability, while preserving the irreplaceable role of human judgment, intuition, and critical thinking in the design process.

The integration of artificial intelligence tools in the studio was designed to complement rather than replace conventional design methods. Therefore, AI-driven platforms were introduced alongside commonly used digital tools and traditional design practices such as sketching, diagramming, and conceptual modeling. The objective of this approach was to provide students with advanced technological resources that could enhance design exploration, analysis, and visualization while preserving the essential role of human judgment, creativity, intuition, and critical thinking within the architectural design process.

To ensure consistency and methodological clarity, a set of selection criteria was established for identifying the AI tools used in this study. The selection process was guided by several considerations. First, the tools needed to be relevant to architectural design activities, meaning that they could support tasks such as data analysis, concept generation, environmental analysis, spatial organization, visualization, or presentation. Second, the selected tools had to be accessible and easy to learn, ensuring that students could use them effectively within the limited duration of the studio experiment.

Third, preference was given to tools that allow rapid generation of design alternatives, enabling students to explore multiple design options within a short period of time. Fourth, the tools needed to support different stages of the design process, ensuring that AI integration could be evaluated across the entire workflow rather than being limited to a single design phase. Finally, the tools were selected based on their availability as online platforms or widely accessible software, which reflects the types of tools that students can realistically adopt in educational environments.

Figure 1 presents samples of interfaces of selected AI tools that were used in study during the six stages of the design process.

Figure 2 presents selected student design outputs, showing earlier work using traditional methods alongside outputs produced during the case study with AI assistance, reflecting the progression of the architectural design process from the pre-design stage to the final presentation. The displayed materials include site analysis, initial sketches, developed design concepts, design development drawings, design documentation supported by structural systems, and final project presentations. These outputs demonstrate how AI-supported tools were integrated across different stages of the studio.

5. Results

This section presents the findings of the study on the impact of AI-assisted tools in architectural design education. The results were analyzed across the six project stages using the six evaluation indicators. Data were collected from multiple sources, including student surveys, faculty assessments, and final project reviews, and were measured using a five-point Likert scale. The analysis includes mean scores, standard deviations, and comparative assessments with prior studies to provide a comprehensive understanding of AI’s influence on both student performance and perception throughout the design process.

As shown in Figure 3, in the Pre-design stage, efficiency (4.2) and adoptability (4.4) were high, confirming AI’s effectiveness in site analysis, data processing, and early contextual understanding. Creativity (3.6) and accuracy (3.5) remained moderate, indicating that AI serves mainly a supportive role at this stage. The Conceptual Design stage recorded the highest creativity (4.8), along with very high efficiency (4.6) and AI integration (4.7), confirming AI’s strong influence in early ideation and form generation, as predicted in the theoretical framework.

In the Schematic Design phase, accuracy (4.4) and environmental impact (4.5) improved, reflecting AI’s shift toward performance-based evaluation, while creativity (4.0) was moderate, showing functional and environmental constraints increasingly guide design decisions. During Design Development, efficiency (4.5) and accuracy (4.6) remained high, emphasizing AI’s role in workflow optimization, coordination, and technical precision. Creativity (3.9) slightly decreased, suggesting innovation is now focused on materials, systems, and technical solutions rather than overall form. The Design Documentation stage showed the lowest scores: efficiency (3.4), creativity (2.7), and AI integration (3.0), indicating students struggled to effectively apply AI at this technical stage. Manual drafting and instructor guidance remained essential. Finally, the Presentation and Final Critique stage achieved peak performance, with efficiency (4.8), creativity (4.6), AI integration (4.7), and adoptability (4.8), confirming that AI visualization, rendering, and storytelling tools significantly enhance design communication and visual impact.

The results of mean and standard deviation in (Figure 4), indicate clear variations in the impact of AI across the different design stages. The highest overall mean score was recorded at the Presentation and Final Critique stage (M = 4.55, SD = 0.31), highlighting the strong role of AI in visual communication, rendering, and final presentation quality. This was closely followed by the Conceptual Design stage (M = 4.47, SD = 0.37), where AI significantly supported idea generation and creative exploration. The Design Development stage also showed a high mean value (M = 4.32, SD = 0.25), indicating effective use of AI in refining and technically developing design solutions. The Schematic Design stage demonstrated a similarly positive impact (M = 4.20, SD = 0.24), reflecting stable AI support during mid-level design decisions. In contrast, the Pre-design stage recorded a slightly lower mean (M = 3.93, SD = 0.34), as AI was mainly used for preliminary analysis and early exploration. The lowest mean score was observed in the Design Documentation stage (M = 3.20, SD = 0.37), confirming that AI has limited effectiveness in highly technical drafting and construction documentation tasks. Overall, the relatively low standard deviation values across all stages indicate a high level of consistency in participants’ evaluations.

6. Discussion

In the Pre-design phase, students reported high usefulness of AI for organizing information, conducting research, and generating preliminary ideas. Survey results showed high Likert ratings for efficiency and adoptability, which faculty feedback and project reviews confirmed. Figure 2 shows a mean of 4.2 (SD = 0.6), indicating consistent performance and suggesting that AI effectively supports early-stage planning and conceptual clarity.

During the conceptual design phase, perceptions were more varied. Students valued AI for creativity and idea exploration, though some questioned its ability to fully support conceptual innovation. Faculty noted differences in originality and coherence. Figure 2 shows a mean of 3.9 (SD = 0.7), reflecting variability, particularly in creativity, highlighting that AI aids visualization but cannot replace individual creative thinking.

In the design development phase, AI improved workflow efficiency and technical detailing, with high Likert scores for efficiency and ease of use. Some students faced challenges integrating complex elements. Faculty observed moderate variation in performance. Figure 4 shows a mean of 4.0 (SD = 0.8), emphasizing that AI supports process efficiency but cannot replace technical skill or judgment.

The design documentation phase showed the greatest variability. Students reported lower scores for documentation quality and adoptability, indicating difficulties in producing comprehensive records. Faculty reviews confirmed these challenges. Figure 2 shows the lowest mean of 3.4 (SD = 0.9), particularly affecting documentation and presentation, demonstrating that AI alone cannot ensure consistent quality.

Finally, in the presentation and final critique phase, students found AI highly useful for visualization, rendering, and preparation. Faculty observed consistent presentation quality. Figure 4 shows a mean of 4.1 (SD = 0.5), confirming AI’s effectiveness in enhancing visual communication and final presentation outcomes.

The findings of this study reveal a differentiated impact of AI integration across the architectural design process, highlighting that its effectiveness is not uniform but strongly stage-dependent. The conceptual design and schematic design stages emerged as the most positively influenced phases, particularly in terms of creativity, visualization, and the ability to rapidly explore multiple design alternatives. This can be attributed to the generative and iterative nature of AI tools, which align closely with the exploratory character of early design thinking. These results are consistent with recent research in architectural education, such as the work of Fatmanur Atalay, which emphasizes the role of AI in enhancing idea generation, supporting design exploration, and enabling students to efficiently evaluate multiple design scenarios [8].

In contrast, the design documentation stage demonstrated the lowest level of improvement, reflecting the inherent limitations of AI in tasks requiring high levels of technical accuracy, standardization, and detailed construction knowledge. At this stage, students relied more heavily on manual skills, professional conventions, and instructor guidance, indicating that AI currently functions as a supportive rather than a substitutive tool in technically demanding phases. This divergence across stages underscores the need for a structured and pedagogically guided integration of AI within architectural education. While AI significantly enhances creativity and efficiency in early and presentation stages, its role in later technical phases remains constrained, requiring critical human oversight to ensure accuracy and reliability. Overall, these findings reinforce the argument that AI should be positioned not as a universal solution, but as a context-sensitive tool whose value depends on the nature of the design task and the stage of the process.

Comparative Matrix Analysis: Literature vs. Experimental Results

To better understand the results, the outcomes of the design studio experiment were added to be three-dimensional matrix, as shown in

Table 5. Three-dimensional matrix based on experimental results (source: Authors).

		Measurement Indicators
		Efficiency	Creativity	Accuracy	AI Integration	Adoptability	Environmental/Architectural Impact
Stages of Design Process	Pre-design	AI helps during data gathering and contextual analysis, allowing designers to make more informed decisions in the early stages of the project.	AI-generated concept visuals can help spark early design ideas, but they still need human refinement to achieve coherence and convey a clear design intention.	AI can improve site understanding and analysis, but its results still need to be checked and verified by humans to ensure accuracy.	AI tools were smoothly incorporated into analytical tasks, improving both efficiency and clarity.	Students found both text- and image-based AI tools to be intuitive and easy to use throughout the design process.	AI and GIS tools helped with early environmental and contextual analysis, enhancing students’ understanding of the site and the relevance of their design decisions.	Results according to five-point Likert scale
		4.2	3.6	3.5	4.0	4.4	3.9
	Conceptual Design	AI sped up the ideation process and made it easier to quickly generate design forms.	AI tools fostered creative exploration and supported the generation of multiple conceptual alternatives.	Maintaining conceptual alignment still relied on instructor feedback to ensure clarity and coherence in the design.	A high level of integration was achieved, with AI tools effectively supporting digital sketching and early-stage visualization.	Students reported high levels of enjoyment and engagement when working with generative visual AI tools.	Generative AI tools promoted sustainable form-finding and encouraged the exploration of passive design strategies.
		4.6	4.8	3.8	4.7	4.6	4.3
	Schematic Design	AI facilitated efficient testing of space layouts and assessment of functional performance.	Creativity was guided by AI-generated performance feedback, which helped refine design decisions.	AI improved spatial accuracy while also supporting more informed environmental considerations in the design process.	The effective integration of AI depended on students’ technical skills and their familiarity with the tools.	Successful AI integration relied on students’ technical skills and their comfort in using the tools.	Energy simulation tools improved the designs’ lighting and ventilation performance.
		4.3	4.0	4.4	4.1	3.9	4.5
	Design Development	AI helped speed up the detailing process and enabled smoother integration of structural logic.	Innovation was most evident in the selection of materials and the design of building systems.	AI enhanced consistency in drawings and helped coordinate different design elements more effectively.	AI integration was feasible, but its effective use required guidance from instructors.	Students adapted quickly and effectively to using generative visual AI tools.	AI-assisted optimization of structures and materials contributed to reducing construction waste.
		4.5	3.9	4.6	4.3	4.2	4.4
	Design Documentation	The partial automation of tasks helped improve overall workflow efficiency.	At this stage, the focus was more on technical precision than on creative exploration.	AI-assisted clash detection proved useful, but manual drafting remained the primary method.	Integration was limited, with AI use mostly remaining at a basic level.	Students faced challenges in effectively applying AI tools for documentation tasks.	AI played a moderate role in addressing environmental considerations.
		3.4	2.7	3.8	3.0	3.1	3.2
	Presentation and Final Critique	AI significantly reduced the time needed to prepare presentations.	AI tools enhanced storytelling, rendering, and video-based design communication.	Visual outputs still required manual editing and instructor guidance to ensure accuracy.	AI tools played a key role in producing the final project deliverables.	Students enthusiastically embraced AI-based visualization tools throughout the design process.	AI visualization tools improved the representation of environmental factors and enhanced spatial perception.
		4.8	4.6	4.0	4.7	4.8	4.4

, and was reviewed alongside the initial analytical matrix that had been developed from earlier studies (Table 1). This step helped reveal the points where the theoretical expectations matched what actually happened in the studio and where they did not. Through this comparison, the study offers a clearer and more grounded view of how AI functions not only as a theoretical concept, but also as a practical and experiential tool that shapes students’ learning throughout the architectural design process.

The comparative analysis between the literature-based expectations and the experimental results reveals both strong convergence and notable deviations in the role of artificial intelligence (AI) across the architectural design studio stages. Overall, the findings confirm that AI is most influential during the early conceptual and final presentation phases, while its impact remains comparatively limited during the technical documentation stage.

The proposed three-dimensional matrix is structured along three axes:

(1)

Design stages;

(2)

Evaluation indicators;

(3)

Data sources (student surveys, faculty assessment, and project evaluation).

This structure allows for a multi-layered analysis of AI impact across both process and performance dimensions.

• Efficiency

Previous studies emphasized that AI significantly enhances workflow efficiency, particularly in pre-design, conceptual, and documentation stages through automation, simulation, and BIM-based processes [19,23]. The experimental results strongly support this assumption. High efficiency scores were recorded during Conceptual Design (M = 4.6) and Presentation & Final Critique (M = 4.8), indicating that students benefited greatly from AI in accelerating ideation, visualization, and project delivery. However, a noticeable decline in efficiency appeared during the Design Documentation stage (M = 3.4), suggesting that students still struggle to fully exploit AI tools for highly technical drafting tasks, despite the strong emphasis on automation in the literature.

• Creativity

The literature consistently highlights the creative potential of generative AI, particularly in the conceptual phase [28]. This expectation was strongly validated by the experimental findings, where Conceptual Design recorded the highest creativity score (M = 4.8), followed closely by Presentation & Final Critique (M = 4.6). These results demonstrate that AI not only supports idea generation but also enhances architectural storytelling and visual expression. In contrast, creativity dropped sharply during Design Documentation (M = 2.7), aligning with previous studies that describe this phase as execution-oriented rather than exploratory [22].

• Accuracy

The literature suggests that AI improves analytical accuracy but still requires human verification [23,32]. This position is strongly reflected in the experimental data. Accuracy steadily increased from Pre-design (M = 3.5) to Design Development (M = 4.6), indicating that AI is more reliable when design information becomes structured and technically defined. During Presentation & Final Critique (M = 4.0), accuracy remained high but continued to depend on manual refinement, reinforcing the literature’s emphasis on the complementary role of human judgment.

• AI Integration

The literature describes AI integration as stage-dependent, with higher levels during generative and simulation phases and lower levels during documentation [19]. The experimental findings confirm this pattern. High integration was observed in Conceptual Design and Final Presentation (both M = 4.7), where AI supported visualization, form generation, and digital communication. Conversely, integration dropped during Design Documentation (M = 3.0), demonstrating that AI adoption in technical documentation is still limited in educational practice despite its theoretical potential.

• Adoptability

While earlier studies predicted that AI adoptability would depend heavily on user expertise and training, the experimental data revealed consistently high levels of student acceptance, particularly in Pre-design (M = 4.4) and Final Presentation (M = 4.8) stages. This indicates a higher level of confidence and openness toward AI tools among students than what was previously anticipated in the literature. The lowest adoptability score was again recorded in Design Documentation (M = 3.1), reflecting practical challenges in applying AI to rigid technical workflows.

• Environmental and Architectural Impact

The literature characterizes AI’s role in environmental performance as emerging and primarily analytical [26]. The experimental results support this view while also indicating moderate progress. Environmental impact scores were relatively high in Schematic Design (M = 4.5), Design Development (M = 4.4), and Final Presentation (M = 4.4), where simulation, optimization, and visualization tools were actively used. However, the lower value in Design Documentation (M = 3.2) suggests that environmental intelligence remains underutilized during final technical production.

To enhance the methodological rigor of the study and ensure the validity of the findings, a set of statistical analyses was conducted, including Spearman’s rank correlation coefficient, Cronbach’s alpha, and the Friedman test. These non-parametric methods were deemed appropriate given the ordinal nature of the Likert-scale data and the relatively small sample size (n = 17).

Spearman’s rank correlation coefficient was used to explore the relationships between the six evaluation indicators (efficiency, creativity, accuracy, adoptability, environmental impact, and AI integration). The results indicated the presence of moderate correlations between certain indicators, particularly between efficiency and adoptability, as well as between creativity and AI integration. Importantly, no excessively high correlations were observed, which suggests that each indicator captures a distinct yet related dimension of AI’s impact on the architectural design process.

To assess the reliability of the evaluation framework, Cronbach’s alpha was calculated, yielding a value of approximately α ≈ 0.78. This indicates an acceptable level of internal consistency, confirming that the selected indicators collectively provide a reliable and coherent measure of performance while maintaining conceptual independence among variables.

Furthermore, the Friedman test was employed to examine differences across the six design stages. The results revealed statistically significant variations (p < 0.05), demonstrating that the influence of AI is not uniform but varies across different phases of the design process. This finding reinforces the idea that AI plays a more prominent role in certain stages, particularly those associated with conceptual exploration and visual communication.

The ranking derived from the Friedman analysis is consistent with the descriptive findings, where the Presentation and Final Critique stage demonstrated the highest level of AI impact, followed by the Conceptual Design stage, while the Design Documentation stage recorded the lowest impact. These results highlight that AI tools are most effective in visually oriented and ideation-driven stages, whereas their influence is more limited in highly technical and documentation-intensive phases.

In general, the experimental findings largely validate the trends reported in previous studies, particularly regarding the strong role of AI in enhancing efficiency, creativity, and visualization in early and final project stages as shown in

Table 6. Evaluation of AI Impact on student design performance: expected vs. observed outcomes (source: Authors).

		Expected Effect Based on the Literature	Observed Effect (Experimental Results)	Interpretation
Indicators	Efficiency	High—AI expected to enhance productivity and workflow	High—Students reported improved time management and faster iteration	Confirms literature; AI supports efficiency in early stages
	Creativity	High—AI anticipated to expand design exploration	High—Participants found AI to significantly inspire idea generation	Experimental results exceed expectations, showing deeper creative engagement
	Accuracy	Moderate—AI tools support analysis but depend on user skill	Moderate—Precision improved slightly but still relied on human judgment	Consistent with literature; AI remains a supportive analytical tool
	AI integration	Increasing—Adoption expected to grow with accessibility	High—Students integrated AI widely in early and final stages	Shows faster adoption than anticipated in literature
	Adoptability	Variable—Predicted to depend on user experience	High—Students showed strong willingness to continue using AI	Surpasses earlier predictions; indicates growing confidence with AI
	Environmental and architectural impact	Emerging—AI expected to have limited environmental influence	Moderate—Some progress in contextual analysis but still developing	Partial improvement: AI use in environmental analysis remains early-stage

. However, the results also reveal a clear gap between theoretical expectations and practical implementation, especially in the Design Documentation stage, where AI’s technical potential is not yet fully realized in studio-based education. Furthermore, the high levels of adoptability and integration observed among students indicate that AI is being embraced more rapidly in academic environments than anticipated in the literature, suggesting a shift in digital design culture within architectural education.

The findings of this study extend beyond identifying the differential impact of AI across design stages and point toward the necessity of a structured human–AI collaboration framework within architectural studios. The results demonstrate that AI contributes most effectively during early ideation and visualization phases, where it enhances exploratory thinking, rapid iteration, and alternative generation. In contrast, later technical stages continue to rely on human judgment, critical reasoning, and professional responsibility.

This stage-sensitive distribution of influence suggests that AI integration should not be uniform across the design process but strategically aligned with pedagogical objectives and cognitive demands at each phase. Rather than positioning AI as a replacement for human creativity, the evidence supports a complementary partnership model in which AI augments analytical capacity and generative exploration, while human designers retain authorship, ethical responsibility, and critical evaluation.

Accordingly, the study advocates for a structured integration framework grounded in three principles: calibrated stage-based application, faculty-guided critical mediation, and ethical awareness embedded within studio culture. Such a framework enables architectural education to transition toward an intelligent yet human-centered paradigm, ensuring that AI functions as a cognitive collaborator that strengthens, rather than diminishes, reflective design practice.

Recent advancements in AI integration within the built environment extend beyond educational applications toward hybrid modeling frameworks that combine analytical, numerical, and data-driven approaches. For instance, recent studies demonstrate how AI techniques can be integrated with physics-based models and experimental validation to enhance both the reliability and interpretability of design systems. Such approaches highlight the potential of AI not merely as a generative or assistive tool, but as part of a comprehensive decision-support framework that bridges simulation, performance evaluation, and design optimization. This perspective suggests that while the current study focuses on educational studio contexts, future developments could extend AI integration toward more performance-driven and professionally applicable architectural design processes [2].

7. Conclusions

This study investigated the integration of artificial intelligence (AI) within architectural design studios and its impact across the six key stages of the design process. The findings indicate that AI had the most significant influence during the early and expressive stages (pre-design, conceptual design, and presentation) where it enhanced creativity, efficiency, and innovation. Its impact was more limited during the technical stages, such as design development and documentation, highlighting the need for better alignment between AI tools and advanced design skills.

Both students and faculty members acknowledged AI’s value in education, though from different perspectives. Students highlighted its creative and time-saving benefits, while faculty emphasized its analytical and reflective potential. These complementary viewpoints suggest that AI is most effective when used to support, rather than replace, human creativity and critical judgment.

The study concludes that a balanced and structured approach is essential to fully leverage AI’s educational benefits in design learning. When applied thoughtfully, AI can serve as both a creative partner that expands imagination and a pedagogical assistant that supports reflection and precision. This research contributes to the ongoing discussion on digital transformation in architectural education and provides a foundation for fostering a more adaptive, ethical, and human-centered design pedagogy in the age of artificial intelligence. The findings should be interpreted within the context of architectural education and not overgeneralized to professional practice without further validation.