Rapid Continuous Improvement Platform through Generative AI: A feasibility study of Automation in Construction Cost Budgeting (RACIP-COST)

英国的建筑微型和中小型企业已经被低生产率和缺乏增长所淹没。只有27%的中小企业在2022年增长（Satista, 2023）。此外，英国建筑业中小企业面临增长所需的劳动力短缺问题。商业、能源和工业战略部（2023年）和人工智能办公室（2023年）都主张通过人工智能在英国发展中小企业。根据这些政府政策，人工智能需要通过不断提高建筑成本管理的生产力来支持增长。建筑业中小企业面临着成本和间接费用等非增值活动的财务问题，例如没有自动化的员工工作时间过长。生成式人工智能是一种机器学习工具，能够减少成本管理所需的工作时间。此外，通过自动化引导的持续改进工具，例如文本到文本输出或图形到文本的生成人工智能，将大大减少建筑和工程项目的管理费用。施工预算过程导致了工程量清单（BOQ）的产生。这个过程需要直接和准确地量化建筑或工程项目所需的建筑材料和任务。作为这个建设预算过程的一部分，个别费率乘以每个描述的数量，以产生详细的成本。boq的准确性被批评为通过纠正措施导致成本超支和生产力问题的主要原因（Love等人，2018；Omotayo等人，2022）。这些成本预算工作很乏味，而且会产生相当多的系统性错误。在过去的十年中，出现了一些计算机辅助设计（CAD）测量和预算软件，例如BlueBeam， RIB CostX， REVIT和PlainSwift。这些与第五维建筑信息模型（5D BIM）相关的软件应用程序面临的挑战是，每年的许可和订阅费用高昂，制作清单所需的技术知识以及定期培训成本和要求。一个以软件应用程序形式的生成式人工智能平台将支持建筑中小企业产生更准确的成本，减少错误并产生持续改进的方法。因此，本可行性研究旨在通过原型软件，论证建筑业中小企业如何通过成本预算的持续改进，创造出一条成长之路。RACIP-COST应用程序是通过机器学习在其他领域的更大应用程序进行持续改进，例如绩效测量，采购和控制。

Construction micro-small and medium enterprises in the UK have been inundated with low productivity and a lack of growth. Only 27% of SMEs grew in 2022 (Satista, 2023). Additionally, construction SMEs in the UK experience labour shortages required for their growth. The Department for Business, Energy and Industrial Strategy (2023) and the Office for Artificial Intelligence (2023) both advocated for SME growth in the UK through AI. In line with these government policies, AI is needed to support growth through productivity enhancements in construction cost management continuously. Construction SMEs experience financial issues for cost, and overheads hinged on non-value-adding activities, such as excessive staff hours without automation. Generative AI is a machine learning tool capable of reducing staff hours required for cost management. Furthermore, a continuous improvement tool guided through automation, such as Generative AI in text-to-text outputs or graphics-to-text, will considerably reduce overheads in construction and engineering projects. The construction budgeting process leads to the production of bills of quantities (BOQ). This process necessitates directly and accurately quantifying construction materials and tasks required for a construction or engineering project. As part of this construction budgeting process, the individual rates are multiplied against each described quantity to produce detailed costs. The accuracy of BOQs has been critiqued as the leading cause of cost overrun and productivity issues through corrective measures (Love et al., 2018; Omotayo et al., 2022). These cost-budgeting tasks are tedious and produce considerable systematic errors. In the last decade, there has been an emergence of computer-aided design (CAD) measurement and budgeting software such as BlueBeam, RIB CostX, REVIT, and PlainSwift, to mention a few. The challenge with these software applications that are now related to the fifth dimension of Building Information Modelling (5D BIM) is the high cost of annual licensing and subscriptions, the technical knowledge required to produce BOQs and regular training costs and requirements.A Generative AI platform in the form of a software application will support construction SMEs in producing more accurate costs, reducing errors and producing an approach for continuous improvement. Therefore, through prototype software, this feasibility study aims to demonstrate how construction SMEs can create a path for growth through continuous improvement via cost budgeting. The RACIP-COST application is about continuous improvement through machine learning for a much larger application in other sectors, such as Performance measurement, procurement and control.