人工智慧於我國政府服務之應用分析 ── 以2018-2022政府服務獎為例

Abstract

本研究欲探討目前我國政府實務上應用人工智慧科技的實際應用情況，並提 出相關政策建議，提供未來研究或是政府服務之參考。 本研究選擇我國公部門最具代表性的評價和獎勵機制—「政府服務獎」當作 研究案例，並將公部門應用人工智慧的過程分為三大研究面向:從最初導入應用 的「動機與關鍵影響因素」，到後續應用中，產生何種「公共價值」以及相關的 「風險、挑戰與課責」。從第一屆至第五屆政府服務獎得獎案例中，挑選涵蓋人 工智慧應用的案例，以個案分析的方式深入進行訪談進行探究，同時輔以資訊專 家的建議。 研究發現，在實務應用中，資料治理成熟度對於人工智慧應用至關重要;人 工智慧科技採用的動機與是否獲得官員、公務員的支持密切相關;而在政府服務 中導入人工智慧科技創新時，由上而下的政策指引、領導推動和支持也是至關重 要的影響因素。此外，人工智慧能有效促進理性決策優化業務決策，並促進「資 料的跨域加值」、以及「業務的跨域協作溝通」，從而增進公共價值，提高人民對 政府的信任度和生活品質。在風險、挑戰與課責方面，政府在人工智慧應用中面 臨資訊安全、數位落差等風險問題。責任歸屬方面，目前實務上應用都是 「human in the loop」的決策模式，也就是所有的責任歸屬仍然在於「人」，不在 「機器」。

This study aims to investigate the current practical application of artificial intelligence (AI) technology within Taiwan's government and to provide relevant policy recommendations for future research or government services. The research selects the "Government Service Award"—Taiwan's most representative evaluation and reward in the public sector—as the case study. The process of AI application in the public sector is divided into three major research dimensions: the "motivation and key influencing factors" for initial implementation, the "public value" generated during subsequent application, and the associated "risks, challenges, and accountability." The study conducts in-depth interviews with selected cases from the first to the fifth Government Service Award winners that encompass AI applications, supplemented by expert advice from information specialists. The findings indicate that in practice, the maturity of data governance is positively correlated with the application of AI in government. The motivation for adopting AI technology is also closely related to whether it gains the support of civil servants. Additionally, top-down policy guidance, leadership promotion, and support are critical influencing factors when introducing AI technology innovations in government services. Furthermore, AI can effectively enhance rational decision-making, and promote "cross- domain data value addition" and "cross-domain task collaboration and communication," thereby increasing public value, improving people's trust in the government, and enhancing quality of life. In terms of risks, challenges, and accountability, the government faces issues such as information security and the digital divide in AI applications. Regarding accountability, the current practical applications of AI in government still follow a "human-in-the-loop" decision-making model, meaning that all accountability remains with the "human" and not the "machine."