AI技術作為新聞邊界物： 台灣醫藥記者勞動與專業實踐的質性研究

Abstract

生成式人工智慧（AI）的快速發展，重塑全球新聞產業的勞動常態與專業邊界。本研究聚焦於具備高度專業知識門檻、低容錯率且肩負重大公共衛生責任的「醫藥新聞」領域，探討台灣醫藥記者在面對AI技術浪潮時，如何感知、運用與調適這項新興工具。本研究運用可供性理論（Theory of Affordances）與邊界物理論（Boundary Object Theory）作為分析理論，探究AI如何改變新聞產製流程、重構勞動關係，並挑戰既有的新聞專業價值與倫理規範。

本研究採用質性研究方法，針對12位涵蓋電視、網路、報紙、雜誌及自由接案的台灣線上醫藥記者進行深度訪談，了解新聞產製流程上，AI實質嵌入前、中、後端工作，觀察到記者從單純的內容生產者轉型為人機協作管理者，看似AI賦能醫藥記者，提升形式上效率，卻同時衍生出不可見勞動，像是額外投入精力學習AI運用外，還得把關AI提供所有資訊，這也讓新聞價值與倫理上產生劇烈性改變，新聞工作正經歷媒介化重構。

總結而言，AI技術不僅是新聞編輯室的輔助工具，更是重塑新聞專業邊界的動態基礎設施。面對AI時代，醫藥記者不單只能靠個人，媒體組織應提供系統性培訓，並制定動態更新的AI使用指引。本研究從不同類別和層級第一線醫藥記者的視角，來檢視新科技帶來的衝擊，可能因應產生的問題，希望為台灣傳播實務與公衛風險溝通提供具體參考。

The rapid proliferation of generative artificial intelligence (AI) is fundamentally restructuring labor practices and professional boundaries across the global news ecology. This study investigates the domain of health and medical journalism—a field characterized by highly specialized knowledge, stringent accuracy requirements, and critical public health imperatives—to examine how Taiwanese health journalistsperceive, appropriate, and adapt to emerging AI technologies. Grounded in the Theory of Affordance and Boundary Object Theory, this research conceptualizes AI as an intermediary object to explore how its technological materiality reconfigures news production workflows, alters labor relations, and challenges established journalistic values and ethical paradigms.

Utilizing a qualitative approach, in-depth interviews were conducted with 12 Taiwanese health and medical journalists spanning television, digital media, print, magazines, and freelance sectors. The empirical findings indicate that AI has become deeply embedded across the pre-production, production, and post-production phases of news work, catalyzing a role transition for journalists from primary content creators to managers of human–AI collaboration. Crucially, while AI enhances operational efficiency, it simultaneously creates new dimensions of "invisible labor," compelling journalists to invest significant cognitive and temporal resources into prompt engineering, systematic verification, and algorithmic oversight. These shifts signify a broader process of mediatized restructuring within the profession. This study concludes that AI operates not merely as an auxiliary tool, but as a dynamic infrastructure reideating the boundaries of journalistic professionalism.

Consequently, adaptation cannot remain an individual burden; media organizations must institute systematic training and iterative governance guidelines. By mapping the intersection of technological affordances and institutional practices, this study provides both theoretical advancements and practical insights for contemporary journalism and public health risk communication.