探究國家人工智慧科學、技術與創新之系統性政策工具

Abstract

本研究主要設定聚焦於台灣人工智慧發展做為研究領域，並以「系統性創新」（systemic innovation）理論為基礎，進行相關分析。本研究研析世界人工智慧先進國家（如美國，日本和中國）近年國家政策報告，並歸納各國推動策略及產業面臨困難問題，以系統性創新理論七大面向，綜整出28個系統創新問題。然後，本研究運用DEMATEL based Analytic Network Process（DANP）研究方法，並通過專家問卷，分析探討各系統性創新問題彼此間，所呈現重要影響程度與關連度，並得出影響效益之權重比例，分列優先次序。再者，本研究援引近年來重要國際組織，含聯合國（United Nations, UN）與經濟合作暨發展組織 （Organisation for Economic Co-operation and Development, OECD）所發展，用以評估國家科學、技術、創新政策（Science, Technology and Innovation Policy，STI Policy）之指標，選定十項重要政策工具，對於28個系統性創新問題之解決的重要性，運用灰關聯（Grey Relational Analysis, GRA) 研究方法，探討每個系統性創新問題與每個政策工具之間的關聯和比較，依據所得出重要性排序，提出解決台灣人工智慧發展相關政策工具之優先順序，希望能提供給我國政府推動政策規劃參考。本研究歸納得出研究結論：知識研究和企業創新對台灣人工智慧發展至關重要；政府介入規範並不是台灣現階段產業所需；擴展和市場形成是台灣企業最期盼的政府政策工具及協助措施；經濟性工具和創造市場需求是推動台灣人工智慧發展最重要的政策手段；企業期盼政府能制訂符合AI創新本質及需求的政策工具。最後，本研究期盼所綜整研提的此一整套研究模式，未來能提供我國與其他國家作為相關產業及不同產業之科學、技術、創新政策規劃及政策工具研訂之參考方案。

This research focused on the field in Taiwanese Artificial Intelligence (AI) development and was based on the theory of systematic innovation for analysis. This research organized the industry problems of U.S., Japan and China and explored 28 systematic innovation problems in 7 dimensions. Then this research used DEMATEL based Analytic Network Process (DANP) method, and analyzed these systemic innovation problems, listed the order of ranking according to the impact weight. Furthermore, this research introduced science, technology and innovation (STI) Policy of United Nations (UN) and Organisation for Economic Co-operation and Development (OECD), and elected 10 policy instruments. And then this research used Grey Relational Analysis (GRA) method, explored the correlations and comparisons between each systemic innovation problem and each policy instrument, and finally propose important policy measures for Taiwanese AI development. The conclusions of this research found the following: knowledge research and enterprise innovation are important to AI development. Government intervention is not the primary need at the current stage. Expansion and market formation are the most anticipated assistance for enterprises. Economic instruments and market demand are the most important policy instruments. And the AI development needs more innovative policy instruments in Taiwan. Finally, this research proposed the whole set of research methods and models; the proposed methodology and models can serve as references for defining future STI policies for other Taiwanese industries or industries belonging to other countries.