Experiments using Tetris confirm that people respond sensitively to `` unfair decisions by AI ''
In modern society, there are cases where allocation using algorithms is performed in various situations such as employee task management and part-time shift adjustment. Until now, the impact of such assignments on human relationships was unknown, but experiments using Tetris showed that people were ``unfairly treated'' even if they were assigned by AI. I understand that I care.
The Social Consequences of Machine Allocation Behavior: Fairness, Interpersonal Perceptions and Performance - ScienceDirect
Tetris reveals how people respond to unfair AI
The research was conducted by Mr. Houston Claure of Yale University and Mr. Kim Seyun of Cornell University, Mr. Rene Kyzilczek, and Mr. Marte Jung.
Tetris used in the experiment was created by Mr. Claure and has been modified so that two players cooperate to clear the stage.
In the experiment, Mr. Claure et al. confirmed the reaction by setting the number of turns to the player to 90% more than the partner, 50% equal to the partner, and 10% less than the partner.
As a result of the experiment, it was found that the players who had fewer turns strongly recognized the unfairness that 'the number of turns of their partner is more than their own'.
This result itself was as expected by the researchers, but the researchers were surprised that ``the emotions held were the same regardless of whether the turn assignment was done by humans or by AI.'' .
In addition, it was found that players who got a lot of turns felt that their partner had a low advantage when the assignment was done by the AI. Perceptions of superiority were unaffected when assignments were made by humans.
These effects are what the researchers call 'Machine Allocation Behavior,' similar to the well-established 'Resource Allocation Behavior,' where people It is an observable behavior shown by
Kizilchek hopes that this will lead to more research on the impact of AI decision-making on people, especially when AI makes decisions on an ongoing basis rather than just once.
It should be noted that fairness does not lead to better game play or performance, and evenly assigned turns scored worse on average than unevenly assigned turns. Mr. Claure said, ``The team's performance will be higher if a good player operates many Tetriminos.''
Related Posts:
