本研究透過生成式 AI 協助資料處理，探討資料驅動 (Data-Driven) 的建築計劃工作流程與方法論。
建築計劃在建築專案的前期階段扮演關鍵角色，其目標在於制定設計規範與決策策略，為後續設計階段提供參考價值，並從中挖掘設計議題與提出洞見。此過程依賴系統化的資料處理 (Data Processing)，即針對人類數位經驗 (Human Digital Experience) 的處理，而生成式 AI 可用於處理大量數位經驗，並從中萃取解決方案。
本研究首先針對以下四個領域進行背景說明與文獻探討：(1) 生成式 AI 在建築領域的發展脈絡、(2) 建築計劃學、(3) 資料科學，以及 (4) 資料與數據在建築流程中的應用。有基於此，總結出以生成式 AI 協助資料驅動的建築計劃工作方法與流程。本研究以成功大學理學大樓為示範基地，並依據五個主要步驟驗證方法的可行性：(1) 確認資料收集項目、(2) 大數據的資料探索與分析、(3) 多方案量體生成建模、(4) 方案模擬與評估，以及 (5) 方案決策。
研究的最終成果將系統化地對各階段性步驟的結果進行深入探討與比較，並在此過程中梳理出生成式人工智慧 (GAI) 與人類智慧 (HI) 在各環節中主導與共創的部分，藉此分析新型建築計劃方法所引發的認知變革及工作流程重塑。此外，本研究亦針對所提出模型的通用性進行探討，期望其未來能廣泛應用於不同類型的建築專案。在未來展望部分，本研究基於所建立的資料驅動方法，提出從都市尺度到建築尺度的完整計劃流程圖，涵蓋都市議題的探索、選址分析、房地產評估、商業策略制定等，期望能在實務操作中得以落實與應用。

This research leverages generative AI to facilitate data processing, exploring data-driven workflows and methodologies for architectural programming. Architectural programming is pivotal in the early stages of architectural projects, serving as a foundation for establishing design guidelines and decision-making strategies. These guidelines and strategies are crucial for providing essential reference points in subsequent design phases, as well as for identifying and articulating design issues and insights. The entire process hinges on systematic data processing, which involves the handling and analysis of human digital experiences. Generative AI has the potential to process vast amounts of digital experience data and extract viable solutions, making it an invaluable tool in the architectural programming process.

The increasing complexity of architectural projects, driven by factors such as urbanization, sustainability concerns, and technological advancements, necessitates more sophisticated approaches to programming. Traditional methods, which often rely heavily on qualitative analysis and manual data handling, are increasingly challenged by the sheer volume and diversity of data that contemporary projects generate. This is where generative AI demonstrates its value. By automating the processing of large datasets, generative AI not only reduces the time and effort required for data analysis but also enhances the accuracy and reliability of the insights derived from this data. This ability to quickly and effectively analyze vast amounts of information positions generative AI as a critical enabler of data-driven architectural programming, aligning the architectural discipline with broader trends in data science and computational design.

This study begins with a comprehensive review of the background and literature in four key areas: (1) the development of generative AI within the field of architecture, (2) the discipline of architectural programming, (3) data science, and (4) the application of data and information in architectural workflows. By synthesizing insights from these domains, the research aims to establish a workflow and methodology for data-driven architectural programming, with generative AI playing a critical role in the process. The literature review underscores the significance of integrating AI into architectural practice, highlighting how advancements in AI technologies have already begun to influence various aspects of design and construction. This integration is not just a matter of adopting new tools; it represents a fundamental shift in how architects approach problem-solving and decision-making, moving from intuition-based methods to more empirical, data-driven approaches.

The demonstration site for this research is the Science Building at National Cheng Kung University (NCKU). This site was chosen for its complexity and the richness of data it offers, making it an ideal case study for testing the proposed methodology. The study evaluates the feasibility of the proposed methodology through five key steps: (1) identifying data collection items, (2) exploring and analyzing big data, (3) generating multi-scheme massing models, (4) simulating and evaluating the proposed schemes, and (5) making informed decisions based on the evaluation of these schemes. Each of these steps is designed to systematically integrate generative AI into the architectural programming process, ensuring that AI-generated insights are not only data-driven but also aligned with the creative and contextual requirements of the project.

The final outcomes of this research will systematically explore and compare the results of each of these stages. Throughout this process, particular attention will be given to delineating the roles of generative AI (GAI) and human intelligence (HI) in leading and co-creating various aspects of the workflow. The research seeks to analyze the cognitive shifts and transformations in work processes brought about by this novel approach to architectural programming. The collaboration between GAI and HI is particularly significant because it represents a convergence of machine efficiency and human creativity. While GAI excels at processing data and identifying patterns, HI brings a nuanced understanding of cultural, social, and aesthetic factors that are crucial in architectural design. This symbiotic relationship has the potential to enhance the quality of architectural programs, leading to more innovative and contextually appropriate solutions.

Additionally, the study will address the critical importance of data science in the architectural programming phase. Quantitative data provides objective and factual information, which is essential for ensuring that the content of the architectural program offers meaningful reference value during the design stage. The process of handling and processing this data can be efficiently managed with the assistance of generative AI, which facilitates the extraction and organization of relevant insights. This integration ensures that the architectural program is not only grounded in solid, data-driven foundations but also optimized for subsequent design phases, enabling more informed and precise decision-making. The emphasis on data science reflects a broader shift within the architectural profession towards evidence-based design, where decisions are guided by empirical data rather than solely by experience or intuition.

In the prospects section, this research builds on the data-driven methods established herein, proposing a comprehensive programming workflow that spans from urban scale to architectural scale. This workflow encompasses a range of activities, including the exploration of urban issues, site selection analysis, real estate evaluation, and the formulation of business strategies. The objective is to provide a framework that can be practically implemented and applied in real-world architectural projects. The proposed workflow is not only comprehensive but also adaptable, allowing for customization based on the specific needs and constraints of different projects. This adaptability is crucial for ensuring that the methods developed in this research can be applied across a wide range of architectural contexts, from large-scale urban developments to smaller, more focused projects.

The significance of this research lies not only in its technical contributions but also in its potential to reshape the cognitive processes involved in architectural programming. By integrating GAI into the workflow, the research proposes a new paradigm where AI and human intelligence collaborate, each bringing unique strengths to the table. GAI’s ability to process and analyze large datasets rapidly complements the creative and contextual understanding of human architects, leading to more informed and innovative design decisions. This synergy between AI and HI could redefine the way architects approach the early stages of design, making the process more data-driven and evidence-based. Moreover, this paradigm shift has the potential to democratize architectural design by making sophisticated data analysis tools more accessible to a broader range of practitioners, including those working on smaller projects or in less resource-rich environments.

Looking ahead, the comprehensive workflow proposed by this research could serve as a valuable tool for architects and urban planners. It provides a systematic approach to integrating data at every stage of the programming process, from the initial exploration of urban issues to the final decision-making phase. The inclusion of urban scale considerations, such as site analysis and real estate evaluation, ensures that the programming process is grounded in a broader contextual understanding, thereby enhancing the relevance and effectiveness of the design solutions generated. This broader perspective is essential for addressing the complex challenges that contemporary cities face, from sustainability and resilience to social equity and economic viability.

In conclusion, this research not only demonstrates the feasibility of using generative AI to support data-driven architectural programming but also offers a forward-looking framework that could transform the field. By bridging the gap between AI-driven data analysis and human-led design thinking, the study paves the way for more innovative, efficient, and contextually responsive architectural solutions. The proposed workflow and methodologies have the potential to be adopted and adapted across a wide range of architectural projects, offering a new paradigm for how architects approach the critical early stages of design. Ultimately, this research contributes to the evolution of architectural practice, positioning it at the intersection of technology, data science, and creative design.

1.Brian S. The Program as Problem: Origins and Impact of CRS’s Problem Seeking. ISBN 978-0-935502-95–4, 2015.
2.Cherry E. Programming for Design: From Theory to Practice. John Wiley & Sons, Inc., New York, 1998.
3.Cherry E. Programming for Design: from Theory to Practice. John Wiley & Sons, Inc., New York, 1999.
4.Cherry E. The Architect’s Handbook of Professional Practice, 14th Edition. John Wiley & Sons, Inc., New York, 2008.
5.Duerk D. Architectural Programming: Information Management for Design. John Wiley & Sons, Inc., New York, 1993.
6.Edith Cherry, FAIA, ASLA. The Architect’s Handbook of Professional Practice. American Institute of Architects, Washington D.C., 2008.
7.Hershberger R. Architectural Programming & Predesign Manager. McGraw-Hill, Inc., New York, 1999.
8.Hershberger R. Programming. In The Architect’s Handbook of Professional Practice, 13th edition. AIA, 2000.
9.Kumlin R. Architectural Programming: Creative Techniques for Design Professionals. McGraw-Hill, Inc., New York, 1995.
10.Marshall, C., & Rossman, G. B. Designing qualitative research (Kindle version), 2011.
11.Ma, X., Hummer, D., Golden. Using Visual Exploratory Data Analysis to Facilitate Collaboration and Hypothesis Generation in Cross-Disciplinary Research., ISPRS International Journal of Geo-Information, 6, 2017.
12.Preiser W. Facility Programming: Methods and Applications. Dowden, Hutchinson & Ross, Inc., Stroudsburg, 1990.
13.Pena W., and Parshall S. Problem Seeking: An Architectural Programming Primer, 5th Edition. John Wiley & Sons, Inc., New York, 2012.
14.Sanoff H. Integrating Programming, Evaluation, and Participation in Design: A Theory Z Approach. Ashgate Publishing, Burlington, 1992.
15.Sanoff H. Methods of Architectural Programming. Dowden, Hutchinson & Ross, Inc., Stroudsburg, 1997.
16.邱浩修，後數位化建築演化論-蛻變與湧現，台灣建築學會會刊, 4-6, 2014。
17.陳珍誠，參數化設計：建築設計中的演算思維，建築師雜誌, 510, 74-79, 2017。
18.黃麗鈴，從社會科學研究典範質量之辯探析混合研究，臺灣教育評論月刊, 8（6）, 163–177, 2019。