Using Bots in Strategizing Group Compositions to Improve Decision-Making Processes

摘要

This paper explores utilization of bots created to play as participants in two games that have relevance in the field of economics. The first game is an iterated version of the prisoner's dilemma that has relevance to decisions requiring trust while conducting business and involves a binary decision. The simulation examines one dimension of the Myers-Briggs personality type and its potential association with strategies that may be employed during an iterated version of the game, leading to different levels of performance. A web-based bot is used for such simulation. Results appear to support the possibility of exploiting personality type in this dimension in the context of similar circumstances. The second game is an iterated power to take game that simulates interactions between taxing authorities (takers) and tax-payers (respon-ders), and involves a continuous set of possible decisions for each stage of the game. A specific Myers-Briggs personality type of specified extremity is embedded into each player bot by randomly generating answers to a version of the Myers-Briggs Type Indicator. These answers are individually associated with preferences and strategies that allow the bots to react to the changing game state according to their personality type. This paper explores the combination of, among other things, team sizes, initial conditions of income tax rates in some of the major economies on different continents, adjudication methods for group decisions, as well as personality type extremity. In a large number of simulations, various group compositions in terms of MBTI personality type are matched up against each other under these conditions, and their performance is ranked for each role. Results appear to suggest that selective recruitment practices based on personality type can enhance performance of both takers and responders and may lead to improvements of certain aspects of economic conditions.