光華講壇——社會名流與企業家論壇第6670期
主題:The Emergence of Economic Rationality of GPT(GPT經濟理性的出現)
主講人:清華大學經濟管理學院 劉瀟副教授
主持人:工商管理學院 趙琳教授
時間:11月21日 10:00-11:30
直播平臺及會議ID:騰訊會議 會議ID:873-799-656
主辦單位:工商管理學院 科研處
主講人簡介:
劉瀟, 清華大學經管學院長聘副教授, 2006年本科畢業于中國人民大學, 2012年博士畢業于美國密歇根大學。研究方向為實驗經濟學、行為經濟學、市場設計和信息經濟學。目前擔任三本國際期刊Management Science,Journal of Economic Behavior & Organization,Journal of Behavioral and Experimental Economics的Associate Editor。文章發表在Management Science、Games and Economic Behavior、Journal of Development Economics等期刊。
內容提要:
As large language models (LLMs) like GPT become increasingly prevalent, it is essential that we assess their capabilities beyond language processing. This paper examines the economic rationality of GPT by instructing it to make budgetary decisions in four domains: risk, time, social, and food preferences. We measure economic rationality by assessing the consistency of GPT decisions with utility maximization in classic revealed preference theory. We find that GPT decisions are largely rational in each domain and demonstrate higher rationality scores than those of humans reported in the literature. We also find that the rationality scores are robust to the degree of randomness and demographic settings such as age and gender, but are sensitive to contexts based on the language frames of the choice situations. These results suggest the potential of LLMs to make good decisions and the need to further understand their capabilities, limitations, and underlying mechanisms.
這篇論文旨在評估類似GPT這樣的大型語言模型在語言處理以外領域的經濟合理性,具體通過在風險、時間、社交和食物偏好等四個領域進行預算決策。我們以經典揭示偏好理論中效用最大化為衡量標準,評估GPT的決策在經濟合理性方面的一致性。研究發現,在每個領域,GPT的決策在很大程度上是經濟合理的,并呈現出比文獻報道的人類更高的合理性得分。同時,研究還發現,合理性得分對于隨機性程度和人口統計學設置(如年齡和性別)具有魯棒性,但對于基于選擇情境語言框架的背景則表現出敏感性。這些結果提示了LLMs在做出良好決策方面的潛力,以及深入了解它們的能力、限制和基礎機制的迫切需求。