 |
| July 2026 |
 |
77 |
|
| 專輯論文Special Issue Articles |
| AI「崽崽」與「捏人」實踐：聊天機器人用戶的情動勞動 |
| AI “Zaizai” and “Character-Customizing” Practices: Affective Labor Among Chatbot Users |
|
|
|
 |
(18)
|
| 作者 |
潘舒怡、牟怡 |
| Author |
Shuyi PAN, Yi MOU |
| 關鍵詞 |
人工智能、社交聊天機器人、大語言模型、情動勞動、非物質勞動 |
| Keywords |
artificial intelligence, chatbots, large language models, affective labor, immaterial labor |
| 摘要 |
借助最新的生成式人工智能技術，用戶得以創建智能體。此類AI產品技術上依賴在人機交互中不斷學習演進的大語言模型，用戶和智能體的每一次對話都是對智能體的訓練。借用神話中女媧用黃泥捏人的原型，此過程也被用戶稱為「捏人」。而用戶通過大量對話創建、訓練、引導、重塑智能體的情感締結過程，本質上是一種情動勞動，即勞動者的情感生產以及對生產過程的控制。通過對國內社交類AI聊天機器人用戶為期一年的虛擬民族誌調查，本研究發現，用戶基於「AI崽崽」的民間理論為智能體提供大量的情動勞動，而這種情動勞動則被科技公司剝削用於訓練其大語言模型和留存用戶、掩蓋產品本身的缺陷。本文將情動勞動研究拓展至人機交互場景，並揭示了其與傳統人際社會中情動勞動的不同之處。 |
| Abstract |
By leveraging the latest generative artificial intelligence (AI) technology, users can now create AI agents. Technically, such agents rely on large language models (LLMs) that continuously learn and evolve through human-computer interaction, whereby every conversation with a user serves as training for the agent. Drawing on the mythological archetype of Nüwa creating humans from yellow clay, users also refer to this process as “kneading a person.” The emotional connection process through which users create, train, guide, and reshape an AI agent via extensive conversations is essentially a form of affective labor, that is, the labor that produces or manipulates affects. Based on a one-year virtual ethnographic study of users of domestic social AI chatbots, this research finds that users invest substantial affective labor in AI agents, guided by their folk theories of “AI baby.” This affective labor is then exploited by tech companies to train their LLMs, retain users, and mask inherent flaws in their products. This paper expands the study of affective labor to the context of human-computer interaction and reveals how it differs from affective labor in traditional interpersonal social settings.
本文引用格式：
潘舒怡、牟怡（2026）。〈AI「崽崽」與「捏人」實踐：聊天機器人用戶的情動勞動〉。《傳播與社會學刊》，第77期，頁83–107。
Citation of This Article:
Pan, S., & Mou, Y. (2026). AI “zaizai” and “character-customizing” practices: Affective labor among chatbot users. Communication and Society, 77, 83–107. |
|
|
 |
| No.76 2026 April |
 |
| No.75 2026 January |
 |
| No.74 2025 October |
 |
| No.73 2025 July |
 |
| No.72 2025 April |
|
