Towards a Conceptual Framework for Introducing a Human-AI Collaborative Decision-Support System Model in Architectural Pre-Design
Loading...
URL
Journal Title
Journal ISSN
Volume Title
School of Arts, Design and Architecture |
Master's thesis
Location:
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Authors
Date
2024-12-31
Department
Major/Subject
Architecture
Mcode
Degree programme
Master’s programme in Architecture, Landscape Architecture and Interior Architec
Language
en
Pages
118 + 10
Series
Abstract
This masters thesis presents a data-driven theoretical framework for incorporating AI-driven tools into the early pre-design phase of the architectural design process, aiming to enhance alignment between design objectives and stakeholder expectations and support informed decision-making. Leveraging AI and Machine Learning (ML), this framework promotes a shift in pre-design systems thinking, encouraging a more integrative and dynamic approach, aligning with contemporary challenges and opportunities. During the fast-paced development of digital evolution, understanding how artificial intelligence can integrate into multiple professional domains to enhance human capabilities has become integral. While computational tools are are not new to architecture, there lies a significant gap in utilising AI, particularly for non-generative tasks like data analysis and insight extraction in planning phases. The architectural design process, with its iterative and phase-oriented nature lacks effective methodologies for a human-AI collaborative tool that has potential to support insights in practice, aiding architects in identifying and refining design objectives based on existing data. In order to address these challenges and the increasing complexity of data, this study draws on Architectural Design Processes (ADP), Systems Theory, Decision Support Systems (DSS), and Human-AI Collaboration to develop a framework that introduces a new method for planning in pre-design stage architectural design. The study introduces the application of large language models (LLMs), such as GPT capable of tasks like text extraction, analysis and data preprocessing of existing data, highlighting its role as a insight-extracting tool rather than an optimization model. This approach repositions AI as a tool for analysis in facilitating collaborative workflows, expanding traditional design processes and fostering decision-support systems theory. This is achieved by developing a Human-AI Collaborative Decision-Support Model through a mixed-method approach that combines qualitative analysis with AI-assisted tools. To develop the framework the study examines case studies of three daycare centres in Helsinki, Hopealaakso, Suutarila and Läpinmaki. City of Helsinki documents including architectural guidelines and planning documents were pre-processesed and standardised to ensure consistency in analysis. Key design guidelines were identified and adherence was quantified through metrics such as Weight of Mention (frequency of guideline mentions), Contextual Relevance (depth of guideline integration), and Document Position (placement within critical sections). By leveraging GPT as a multi-purpose large language model the study employed natural language processing (NLP) techniques, including Term Frequency-Inverse Document Frequency (TF-IDF) analysis, to asses the importance of specific guidelines within the documents related to the entire dataset. Clustering and topic modelling methods were applied to categorise and interpret textual data, strengthening the validity of the analysis. Results were mapped through visuals including bar charts and tables to highlight insights and discrepancies of how each guideline was treated among each planning document. Stakeholder feedback was incorporated through informal interviews with architects within the architecture firm AFKS and stakeholders from the city to validate findings and provide iterative refinement of the methodology, ensuring relevancy and applicability. The proposed model introduces a workflow for pre-design data analysis, offering an iterative data-informed approach for developing future planning documents.The analysis identifies gaps within guideline compliance with limited focus in adaptability and user-centred spaces in this specific study. The research emphasises the shift in design thinking, exploring-advocating-encouraging the adoption of Human-AI driven tools for insight analysis and collaborative workflows within the architectural field. By illustrating how AI can enhance human expertise while preserving creativity, the thesis aligns with the principles of the profession while also opening up new avenues for discussion on advancing within its framework. Additionally the framework introduces AI as an analysis and decision-support tool that attempts to align design objectives with stakeholder needs. Future research will involve developing a tool with an user-interface to further facilitate collaboration, building mutual understanding for productive discussions in creating design objectives.Description
Supervisor
Fricker, PiaKeywords
Architecture Pre-Design Phase, hybrid Intelligence, Decision Support Systems (DSS), Human-AI Collaboration, Human-Machine Collective Intelligence (HMCI), stakeholder alignment