STS如何介入AI：ANT取徑朝向AI公共性的思考實驗

人工智能（artificial intelligence, AI）發展對人類未來的影響充滿各種想像，尤其不乏極度樂觀與悲觀者。從科技與社會研究（Science, Technology and Society, STS）觀點，本文重新想像AI、人性與社會，剖析AI政治性並提出拓展AI公共性的可能。本文聚焦於行動者網絡理論（actor-network theory, ANT）的位移取徑，由AI與人性、社會性的位移發展幾個思考實驗：首先跟隨經典的圖靈測試（Turing test）示範的位移工作，說明行動者網絡取徑如何位移諸如「機器是否能思考」與「科技如何影響人類」等問題。其次，本文打開AI的黑箱，說明AI與社會如何相互位移，造成各種政治性，最後，本文提出多元介入促進AI公共性作為另類可能。

關鍵字

人工智能；圖靈測試；行動者網絡理論；位移； AI政治性；科技－社會想像

並列摘要

The question of AI's (Artificial Intelligence) impact on humanity invites a broad range of pessimistic and optimistic views. Drawing on STS insights, this paper examines the relationships between AI, humanity, and sociality by focusing on the politics of AI and AI's potential for publicness. This paper takes an ANT displacement perspective to mobilize thought experiments. First of all, I illustrate the displacements in the classical question, 'Can machines think?' from the Turing test. Second, I propose various interventions to demonstrate how 'AI, humanity and sociality' actor-networks are changing in various displacements. Third, instead of being pessimistic or optimistic, this approach opens up the black box of AI in order to show its current politics. Through these ANT-based STS interventions, I suggest that alternative ways of enacting AI for the public are possible.