與AI科技協作提升工地安全意願之研究

A study on collaboration with AI to improve construction site safety willingnes

建築行業偏高的職業傷害死亡率是全球各國現代化面臨的一項重要議題。由於近期AI科技的進步與成熟，AI安全配備與裝置也逐步在營建工地一一的被導入且越來越普遍。以大語言模型為基礎的互動AI軟體ChatGPT 3.5版本自2022年11月發布以來，就引起大眾媒體和學術界的廣泛關注，爾後2025年1月DeepSeek企業發佈的Deepseek-R1開源模型，其訓練成本低廉更為全球產業帶來AI應用的成長與商機。鑑於此發展趨勢，我們開發一個以科技接受與使用整合理論（UTAUT）及體驗價值為中心的實證模型，來瞭解台灣從業人員對於營建工地使用AI科技（例如智慧安全帽背心、AI影像分析事故辨識、無人機空拍影像巡檢、AI工地安全風險預測模型等）進行安全管理與事故預防之接受與協同意願。為檢驗此模型，我們對台灣營建業242位從業人員進行調查，並採用PLS-SEM統計法進行分析。研究結果表明，AI尊重知覺、AI美學知覺與AI幸福知覺三個體驗價值構念，可以顯著影響台灣營建業員工接受與推薦企業導入工地安全AI。而會促進員工未來與AI協同合作的關鍵因素主要來自於AI效率知覺、AI社群影響、AI幸福知覺與AI推薦意願等。基於這些發現，我們建議營建業者應認真考慮並規劃工地安全AI系統的部署策略，以確保它們被從業人員接受、推薦及協同合作，以高效提升營造工地現場安全。

The high rate of occupational injuries and fatalities represents a critical issue in the AEC industry. With recent maturation of AI technologies, AI-based safety system have increasingly been introduced and widely adopted at construction sites. Since the release of ChatGPT 3.5, a LLM–based interactive AI software, in November 2022, significant attention has been drawn around the world. Subsequently, in January 2025, DeepSeek launched the open-source model DeepSeek-R1, which, due to its low training cost, has further accelerated AI application opportunities and growth across global industries. This study develops an empirical model integrating the UTAUT with experiential value to investigate the willingness of construction practitioners in Taiwan to adopt and collaborate with AI technologies for safety management. Such technologies include smart helmets and vests, AI-based video analysis for accident detection, drone-based aerial safety inspections, and AI-driven predictive models for construction safety risks. To test this model, a survey was conducted with 242 construction practitioners in Taiwan, and the data were analyzed using PLS-SEM. The results indicate that three experiential value constructs (AI esteem perceived, AI aesthetics perceived, and AI spirituality perceived) significantly influence employees’acceptance of and recommendation for AI adoption in construction site safety management. Moreover, the key determinants that foster employees’willingness to collaborate with AI in the future include AI efficiency perceived, AI social influence, AI spirituality perceived, and AI recommendation intention. Based on these findings, this study suggests that construction enterprises should carefully consider and strategically plan the deployment of AI-based safety systems on construction sites, ensuring their acceptance, endorsement, and collaboration by practitioners, thereby effectively enhancing on-site safety in the AEC industry.