人工智慧科研倫理與風險之基本認識

AI Literacy: Ethics and Risks in Research and

生活周遭越來越多事物使用人工智慧（artificial intelligence,以下簡稱AI）技術或相關產品，但人們對於AI科技的看法，往往不是太樂觀天真，就是太悲觀或帶有刻板印象。本文旨在整理與概述AI的倫理議題與風險有哪些面向，內容上主要分成2個部分。第一部分先簡介目前AI系統最常使用的機器學習技術，以及可能涉及的倫理議題與風險。第二部分則以國科會2019年公布的「人工智慧科研發展指引」為例，針對文件中所提及的3項核心價值以及8項行為指引，逐一說明每項理解方式及相關案例；其中8項行為指引又被區分為：任何新興科研都應該遵守的、有關AI技術特性、以及有關AI應用等3個類別。期待本文有助於促進非AI技術人員對於AI倫理與風險的認識。

Artificial intelligence (AI) and its related products are increasingly being integrated into our daily lives. However, people’s perceptions of AI, often influenced by media hype or sci-fi films, range from excessively optimistic and naive to overly pessimistic and even stereotypical. These polarized viewpoints ultimately hinder the development of trustworthy AI, which by definition should be legal, ethically compliant, and technically robust. This paper aims to help non-AI technical professionals understand AIrelated ethical issues and risks by dividing the discussion into two parts. The first part introduces the most commonly used machine learning technologies in AI systems and discusses potential ethical issues and risks. The second delves into the“AI Technology R&D Guidelines,”published by the National Science and Technology Council in Taiwan in 2019. Interpretations of the document’s three core values and eight guidelines are provided. The eight guidelines are categorized based on their relation to emerging technology R&D, AI technology characteristics, and AI application outcomes. Real-world case studies illustrate each guideline, while also highlighting their interplay within each category. In conclusion, this paper offers a foundational understanding of the ethics and risks involved in AI research and development.