时间：2024年5月14日 15：30（周二）

地点：邵逸夫科学馆401

主讲人：许彬

题目：Human-friendly algorithm drives cooperation more

内容摘要：With the development of artificial intelligence (AI), the dynamics of collaboration have evolved from solely interpersonal interactions to include human-machine partnerships. Understanding the factors that influence human-machine cooperation to foster effective relationships is crucial in the AI era. This paper investigates whether algorithms that more closely mimic human behavioral traits are more likely to facilitate human-machine cooperation. Utilizing four fair zero-determinant (ZD) strategies as algorithms, our human-machine interaction experiment revealed that algorithms resembling first-order memory reciprocity better facilitate cooperation. Conversely, algorithms that oppose first-order memory reciprocity, despite offering fair long-term expected benefits, perform poorly in promoting human-machine cooperation. Our numerical simulations further confirm that changes in algorithm scores primarily result from changes in human behavior prompted by alterations in the algorithms, rather than from the changes in the algorithms themselves. This study not only deepens our understanding of ZD strategies but also aids in designing algorithms that align with human values, which is vital for enhancing human-machine cooperation.

专家简介：许彬，女，浙江工商大学经济学院教授，博导，中国行为与实验经济学论坛学术委员会成员。研究领域包括行为经济学、实验经济学、新制度经济学、公共经济学等，同时还进行跨学科研究，如行为经济学与演化博弈的结合，经济学与生态学的结合。代表性成果（合作）主要为在《Scientific Reports》上发表的“Social cycling and conditional responses in the Rock-Paper-Scissors game”和《Nature Communications》上发表的“Extortion can outperform generosity in the iterated prisoner's dilemma”。在行为与实验研究领域发表了10余篇SSCI或SCI文章，并获得省哲学社会科学优秀成果一等奖1次。与生态学家合作进行实证研究，发表多篇ESI高被引文章。主持和参与了国家自然科学基金、国家社科基金以及教育部基金等多个课题。