摘要

系统梳理人工智能(AI)技术,特别是生成式大模型在工程建设领域的最新研究进展与应用挑战。研究表明:以扩散模型与Transformer架构为核心的生成式AI已成为主流技术基座,实现“文本-图像-3D模型”端到端生成,显著提升设计效率;当前研究聚焦多目标优化与可解释性并重、认知交互与感性工学融合,但面临数据版权碎片化、生成结果可解释性不足等问题。面对这些问题,提出企业需平衡模型规模与业务需求、构建高质量知识资产、优化智能体工作流,个人应转型为智能体指挥官、严控AI幻觉、锤炼深度决策与创新能力的建议。研究为行业应对AI驱动的生产力重构提供系统性策略框架,强调“人机协同”而非“机器替代”的核心发展路径。

This research systematically reviews the latest research progress and application challenges of artificial intelligence ( AI) technology, especially generative large models, in the field of engineering construction. Research shows that generative AI, centered on diffusion models and Transformer architectures, has become the mainstream technical foundation, achieving end-to-end generation from “text-image-3D models” and significantly enhancing design efficiency. Current research focuses on the integration of multi-objective optimization and interpretability, as well as the fusion of cognitive interaction and sensibility engineering. However, it faces problems such as fragmented data copyrights and insufficient interpretability of generation results. In response to these issues, this research proposes that enterprises should balance model scale and business needs, build high-quality knowledge assets, and optimize the workflow of intelligent agents; individuals should transform into intelligent agent commanders, strictly control AI hallucinations, and cultivate deep decision-making and innovation capabilities. This research provides a systematic strategic framework for the industry to deal with the productivity reconstruction driven by AI, emphasizing the core development path of “human-machine collaboration” rather than “ machine replacement”.

论文下载链接

预印本下载链接

基金资助: 国家重点研发计划资助项目（2023YFC3804600）； 国家自然科学基金资助项目（52378306）

相关项目资助情况: