October  2023 66
Who is ChatGPT: A College Student, RA or PI?
作者 方可成
Author Kecheng FANG
摘要 本文基於ChatGPT對十幾位傳播學教授提問的回答,借鑒學界和業界的最新研究與討論成果,評估大語言模型在社會科學研究中的三種可能的角色:勤勉的學生、可靠的研究助理、獨當一面的課題負責人。本文進而切入大語言模型的三個關鍵問題:訓練數據、提示語寫作(prompt engineering),以及基於概率統計的模型是否可能具備認知能力。本文最後建議,在ChatGPT橫空出世,顛覆傳統的新形勢之下,社會科學學者當前應該處變不驚,繼續維持自己的主體地位,同時積極尋找將自己的智能與大語言模型的能力相接的方法。
Abstract Based on ChatGPT’s responses to questions from several communication scholars and drawing on the latest research and discussions in academia and industry, I propose three possible roles for large language models (LLMs) in social science research: the diligent student, the reliable research assistant (RA), and the independent principal investigator (PI). I further discuss these roles in relation to three key issues of LLMs: training data, prompt engineering, and whether probabilistic statistical-based models can be cognitively competent. I conclude by suggesting that social scientists should continue to maintain their sovereignty in scholarly endeavors and actively seek ways to tap the intelligence of LLMs to empower their research.


Citation of this article:
Fang, K. (2023). Who is ChatGPT: A college student, RA or PI? Communication and Society, 66, 53–65.
No.67  2024 January
No.66  2023 October
No.65  2023 July
No.64  2023 April
No.63  2023 January
© 2017 Communication & Society Journal. All Rights Reserved.