光華講壇——社會(huì)名流與企業(yè)家論壇第6670期
主題：The Emergence of Economic Rationality of GPT(GPT經(jīng)濟(jì)理性的出現(xiàn))
主講人：清華大學(xué)經(jīng)濟(jì)管理學(xué)院 劉瀟副教授
主持人：工商管理學(xué)院 趙琳教授
時(shí)間：11月21日 10:00-11:30
直播平臺(tái)及會(huì)議ID：騰訊會(huì)議 會(huì)議ID：873-799-656
主辦單位：工商管理學(xué)院 科研處
主講人簡(jiǎn)介：
劉瀟, 清華大學(xué)經(jīng)管學(xué)院長(zhǎng)聘副教授, 2006年本科畢業(yè)于中國(guó)人民大學(xué)， 2012年博士畢業(yè)于美國(guó)密歇根大學(xué)。研究方向?yàn)閷?shí)驗(yàn)經(jīng)濟(jì)學(xué)、行為經(jīng)濟(jì)學(xué)、市場(chǎng)設(shè)計(jì)和信息經(jīng)濟(jì)學(xué)。目前擔(dān)任三本國(guó)際期刊Management Science，Journal of Economic Behavior & Organization，Journal of Behavioral and Experimental Economics的Associate Editor。文章發(fā)表在Management Science、Games and Economic Behavior、Journal of Development Economics等期刊。
內(nèi)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(píng)估類(lèi)似GPT這樣的大型語(yǔ)言模型在語(yǔ)言處理以外領(lǐng)域的經(jīng)濟(jì)合理性，具體通過(guò)在風(fēng)險(xiǎn)、時(shí)間、社交和食物偏好等四個(gè)領(lǐng)域進(jìn)行預(yù)算決策。我們以經(jīng)典揭示偏好理論中效用最大化為衡量標(biāo)準(zhǔn)，評(píng)估GPT的決策在經(jīng)濟(jì)合理性方面的一致性。研究發(fā)現(xiàn)，在每個(gè)領(lǐng)域，GPT的決策在很大程度上是經(jīng)濟(jì)合理的，并呈現(xiàn)出比文獻(xiàn)報(bào)道的人類(lèi)更高的合理性得分。同時(shí)，研究還發(fā)現(xiàn)，合理性得分對(duì)于隨機(jī)性程度和人口統(tǒng)計(jì)學(xué)設(shè)置（如年齡和性別）具有魯棒性，但對(duì)于基于選擇情境語(yǔ)言框架的背景則表現(xiàn)出敏感性。這些結(jié)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(chǔ)機(jī)制的迫切需求。
